Printable View
Hi all
What i would like to know is how the 50D does with birds. I have one and im very unhappy. With my 40D i can shoot at ISO1250 with no problem but with the 50D ISO 1000 is unuseable. the image have just to much noise. Second thin is that the AF is not as good as the 40D.
i bought the camera 3 weeks ago and im unhappy. With still life at ISO 400 and under its super sharp.
Do i have a bad copy or is that the way the camera is...
:confused::confused::(
Any info will be appreciated
Alfred Forns and Artie Morris have a lot of experience with the 50 D and I think they like it a lot. I am sure they will chime in. You might PM them to get the nitty gritty details. However, my understanding is that the AF is spectacular for BIF if you use the center point without the assist points. The noise issue is reduced if you have good exposure control and perhaps use a noise reduction plug in.
In the 40D and 50D a 1250 ISO image is shot at 1600 ISO (overexposed) by the camera and then the raw pixel data values are mathematically reduced by a factor of 5/4 to give you the equivalent of 1250 ISO before saving the file. A 1000 ISO image is actually shot in camera at 800 ISO (underexposed) and then the raw data values are mathematically multiplied by 5/4 to give you the equivalent of 1000. It is true at (almost?) all ISO levels that the higher intermediate ISOs give cleaner images than those shot at lower intermediate values - 160 cleaner than 125, 320 cleaner than 250 and so on. For that reason I only shoot at 100, 200, 400, 800, 1600, 3200 ISO with these cameras. The intermediate ISOs lose you 1/3 stop of dynamic range at one end or other of the tonal scale.
If you are going to compare the cameras then at least compare apples with apples, not apples with bananas. i.e. compare them at the same ISO. Also make sure that your settings for HTP, ALO, NR are comparable between both bodies. I have these things disabled, always, as I only shoot raw.
To save you some time, at a pixel level the 50D does have noisier pixels for any given ISO. I doubt anyone would disagree. However, pixels are for lab techs. Photographers should be more interested in images. When you compare images from each camera, shot at the same ISO, and then look carefully for differences in detail and noise you may find the 50D performs really rather well.
Here are shots from my 40D and 50D at 1600 ISO, processed in Lightroom with default settings and no adjustments. These are crops from a larger frame, designed to simulate shooting a small bird at a distance, and intended for me to determine which camera gives me the better image. I could not care about the pixels.
Please take the time to consider which example wins for detail and which wins for noise, and then check the EXIF to see which is which.
http://lh4.ggpht.com/_4_R8TkwT74w/Sj...70927_1_LR.jpg http://lh6.ggpht.com/_4_R8TkwT74w/Sj...71252_2_LR.jpg
Here are the same two images processed in DPP with default settings....
http://lh4.ggpht.com/_4_R8TkwT74w/Si..._Crop_0005.JPG http://lh3.ggpht.com/_4_R8TkwT74w/Si..._Crop_0005.JPG
I have a full album of results at full stop ISOs from 100-3200 from the 40D, 50D and also my 1D3, here....
From Lightroom - http://picasaweb.google.com/EezyTige...CropToSubject#
From DPP - http://picasaweb.google.com/EezyTige...tCropToSubject
The full images processed from DPP are here - http://picasaweb.google.com/EezyTige...estFullFrames#
If you must compare 100% crops (pixels) then see here - http://picasaweb.google.com/EezyTige...nTest100Crops#
Thx for the fast replys. Like i say i used the 40D and iso 1250 is fine with it so i know how to expose....the problem with the 50D is that at a 1000 ISO the images are unuseable. Its got to much noise and thus loose detail. Af i use centre only but af not as good as 40D.....Im with my hands in my hair
Kobus, I do not understand why you are shooting at 1000 ISO. It is not a good choice, as I already explained. Here are my shots with the 50D at 3200 ISO, processed in DPP with default NR settings. These were not bright conditions, being overcast and 3 stops dimmer than "Sunny 16"....
Full frame image....
http://lh4.ggpht.com/_4_R8TkwT74w/Si..._Full_0006.JPG
Crop to subject....
http://lh3.ggpht.com/_4_R8TkwT74w/Si..._Crop_0006.JPG
100% crop....
http://lh4.ggpht.com/_4_R8TkwT74w/Si...op100_0006.JPG
I am sure that processed with specialised NR software like Neat Image the results would be better.
Can you show us a file that you are unhappy with? If you have a raw original and can host that somewhere for a closer look then that would help determine whether your results are representative.
Great post Tim Thanks for the samples You make valid points and sense !!!!
Currently I'm using both the Mk3 and 50D The 50D has an excellent AF performance and would rate as good as any other Canon. With birds moving straight at you is the best.
Noise wise I'm not having much trouble but have changed my noise reducing software. Now I'm using Toapaz 3 and results are impressive. Personally I would not pick a 40D over the 50D ... more pluses than minuses ... one thing to note If you underexpose and image the noise will be severe even if it only 1/3 !!!!
Cheers, Alfred.
Here's that 3200 ISO "crop to subject" file with NR in Neat Image and some sharpening tweaks in Lightroom. EXIF is intact....
http://lh4.ggpht.com/_4_R8TkwT74w/Sk..._1_Neat_LR.jpg
If I used Photoshop and fine tuned NR and sharpening with layer masks I imagine I could improve on these results, but I don't use Photoshop.
I have a new 50D as well, and it is about to be sold. Although I prefer the 50D b/c of the LCD screen and for birds in flight, the noise is unacceptable. Yes Topaz is pretty good at removing the noise but, frankly, I am happier using my 40D than dealing with it. Aside from that there is a magenta cast to many of my images which I also find annoying. Good luck.
Marina,
- Are you shooting raw or JPEG?
- Do you have HTP enabled?
- Do you have ALO enabled?
- Do you have NR enabled - what strength?
- Do you Expose To The Right, aim for correct exposure in camera, or underexpose
- Do you use intermediate ISOs that are 1/3 stop above the main ISOs or 100, 200, 400 etc.?
- Are you cropping heavily?
- If you shoot raw, which raw software do you use? What is your workflow?
- You mentioned Topaz but what is your usual approach to NR? I'd never heard of Topaz until today.
- Do you have an unedited example file - preferably raw - that you could let me take a look at?
Kobus - Feel your pain, but suggest not throwing it under the bus just yet. I was not a 40D owner, but was surprised with my initial 50D results from my earlier Canon - the 50 had far more to manage than I realized. You likely are much more experienced than I am based on your comments, but here are some things I found as I stuck with it for a few weeks and really found my results improved - part of my issue was aligning the lens (100-400 IS) - the rest of the issue was me learning the camera. If you search 50D lens adjustment, 50D focus, etc. in the data base there are several good strings that helped me. Here are some:
http://www.birdphotographers.net/for...ad.php?t=37621
http://www.birdphotographers.net/for...ad.php?t=29716
http://www.birdphotographers.net/for...ad.php?t=35159
The other was the info Tim Dodd provided above. I did find that I figured out how to get very good images at the high ISO levels - 1600 even in a redwood forest (so shady - not the 50's strongest test, but it performed!). I think Tim suggested posting an image - I got great input to help me figure out where my issues were. Hope you get it sorted, I really like mine now that I know it. BTW - loved birding in your country last summer.
Thx to all for the great replys. I'm just trying to find out if I have a bad copy or not.
Todd I always expose to the right. Sometimes even a stop more. Its not that I always shoot at iso 1000 but sometimes don't have a choice. Thing is I could do it with the 40D and now I can't....
Kobus,
I upgraded from a 40D to a 50D and for most part I was disappointed too, I love the better LCD and the microadjustment but I did not notice any improvement in AI-servo with my own technique and 400 f/5.6 which is center point only and pumping the shutter button when bird is against folaige. It is also noisier than 40D and demands more from the lenses, for example I no longer shoot with my 400 f/5.6 wide open. See my post on 50D AI-servo here- you are not alone :( IMO opinion 50D isn't a real upgrade for 40D owners :D
http://www.birdphotographers.net/for...ad.php?t=37961
I am using NR sw as a standard workflow now and never go above 800, it works as long as you don't try to recover shadows too much. But yes if I had tried it before I would probably have kept my 40D and saved for a MKIV when and if it comes out :)
Kobus, I would love to try to help you but without seeing an unedited example of one of your disappointing images I don't see how I can help.
Please remember, shooting at 1000 ISO is an underexposure within the camera of 1/3 stop, because the camera is actually shooting at 800 ISO. The camera then digitally multiplies the raw data values to simulate shooting at 1000 ISO. This is really little different to shooting at 800 ISO yourself with -1/3 EC dialed in and then boosting the exposure in your editing software.
If you look at the DXOMark ratings between the 40D and 50D there is little to choose between them, but the 50D has significantly higher resolution. There are trade-offs here. It is unlikely that the extra resolution could all be accomplished with no penalty in noise whatsoever. I'm sure we all know by now that Canon's claims for noise where more than a little optimistic where the 50D is concerned and now it appears that the claims only apply to in camera JPEGs, where advanced NR algorithms from the DIGIC 4 procecssor can massage away to worst of it. For raw shooters there is no free lunch here. Here are the comparisons....
http://www.dxomark.com/index.php/eng/Image-Quality-Database/Compare-cameras/(appareil1)/180|0/(appareil2)/267|0/(onglet)/0/(brand)/Canon/(brand2)/Canon
Here's a Q&A session with Chuck Westfall on the 50D regarding noise....
http://forums.dpreview.com/forums/re...ssage=30002197
Please look at the list of questions I asked Marina. If you are using HTP and/or ALO then disable them both. Then see how you get on. And please avoid using ISOs 1/3 stop above the "standard" ISOs. Please also compare images with images, not pixels with pixels.
Based on my own experience with the 50D I do not think the extra pixels should be used to attempt cropping beyond what you would do with the 40D. All that will do is expose the noise more readily, and the extra noise should not come as a surprise. The benefit of the extra pixels should be realised in extra detail captured, not tighter crops. To put it another way, assuming you wanted to produce a 6x4 print at 300ppi you would need 1800x1200 pixels. If you are cropping tighter than that then perhaps you are not close enough to your subject. If you wanted to prepare an 800x533 web image then I would aim to have no less than 1600x1067 pixels after cropping and then resize that to ~50% to achieve the image you want. If you are viewing the files at enlargements greater than that then just consider what that really means. A full 4752x3168 file, viewed at 50% would yield a virtual image of 18x12 on my laptop monitor. That is a pretty hefty enlargement to be scrutinizing from 12" away. Viewed at 100% the virtual image would be 36x24, which is massive when staring at it from 12-18 inches away. On most monitors the virtual image is likely to be even larger than that.
I just hope your expectations are realistic. If you can provide a raw file somewhere then at least I can compare your results with mine.
Excellent points by Tim regarding how to use the extra pixels of the 50D. I think the 50D is a substantial upgrade from the 40D in the AF department when it comes to BIF. I use AI Servo with center point only. Noise can be an issue with the 50D, but can be minimized by nailing your exposure in camera and running NR on the BG. The one thing that I do like with the 50D is the preservation of detail at high ISO. When I use ISOs above 800 I tend to shoot JPEG, because I like the in-camera NR performed by the 50D. Here's a sample of ISO 3200 shot in very low light. I think feather detail is very well preserved. NR was performed on the BG only.
http://birdwhisperer.smugmug.com/pho...87_XeRmt-O.jpg
Here's an ISO 1600 50D image; it's high-key so underexposure (easy to do with a bright BG) would have ruined it.
http://birdwhisperer.smugmug.com/pho...67_tD4MK-O.jpg
And one last image taken at ISO 3200 with the 50D. Light was very low (f/2.8, 1/640, ISO 3200).
http://birdwhisperer.smugmug.com/pho...52_gkSsD-O.jpg
My point with these photographic examples is that the 50D is capable of some pretty impressive stuff at high ISO once you get to know the camera and how to process the images.
Thanks, Doug, for some nice birdy examples. I'm afraid my birdy choices are a bit limited so I've set up another demo with my furry companions. This time I placed them in a shady spot in the garden, lit only from blue sky above. I fired off shots at 400, 800, 1600 and 3200 ISO. The light level in the shade was 5.0 to 5.3 stops dimmer than "Sunny 16", so probably about as dark as you would want it to be for bird photography. e.g. the 400 ISO shot was at 300mm, f/5.6, 1/80 and the 3200 ISO shot was at 300mm, f/5.6, 1/800
I shot raw and processed the files in Lightroom 2.4 with no edits except a WB adjustment. I resized them to 25% on conversion to JPEG, making them 1188x792 pixels, and I have uploaded them as a slideshow album. I've randomised the sequence and I would be interested in whether people feel confident that they can identify which image is which ISO. EXIF is intact so you can look if you want to. The four images play in a continuous loop. The album website will resize the images to fit your monitor if its resolution is too low, so I suggest you press F11 to set your browser to "Full screen" mode and perhaps close any favorites/history sidebar if it gets in the way. Here is the slideshow....
http://picasaweb.google.co.uk/EezyTi...54474951885922
Regardless of which one is the 3200 ISO example I don't see any of them being unuseable as far as IQ/noise is concerned.
Thanks for posting the images Doug I'm at a loss why the difference of opinion and some of the ones having trouble are competent and know what they are doing so they are seeing things !!! Your images and the ones by Tim reflect my results !!
... one thing to note ... if you expect to shoot at high ISO 400 or more and that is under ideal conditions, you will have to deal with noise with software. For better noise performance you can use the Mk3 but at times I like the extra crop factor and use the 50D.
I've got a sneaking suspicion the problem might be to do with heavy cropping, or viewing at huge magnifications (100% is a huge magnification for a 50D file).
Look at it this way. The sensor collects the light. The sensor is the same size in both the 40D and 50D. In simplistic terms they have the same surface area and both collect the same amount of light. Whether you chop the light up into 10 million little squares or 15 million little squares doesn't really affect the total light that can be collected. (I did say this was in simplistic terms)
If we briefly reflect on a full frame camera, such as the 5D2, and its sensor, it has a sensor area about 2.6X greater than that of the little croppers. Provided the glass has the focal length to make use of the extra surface area that sensor will collect 2.6X as much light and thus yield superior noise performance. The size of the sensor is quite significant in terms of noise performance and other aspects of IQ. We all know how terrible compact cameras are at higher ISOs, regardless of the number of pixels. They've always been rubbish.
Back to the 40D and 50D - if we make equal crops from the two cameras, say 1/4 of the frame each then we retain 1/4 of the light captured by the sensor - 25% from each. The playing field is still level. OK, from the 40D we will now only have 2.5 milion pixels and from the 50D around 3.75 million. But we will still have the same area in the crop and the same amount of light (theoretically) to present in our final image.
Now, what happens if you crop from each sensor to an area of, say, 800x533 pixels = 426000 pixels? Well from the 40D that represents just 426000/10,000,000 of the total area = 4.26% of the total sensor area used to capture light. If we make a crop from the 50D, also of 800x533 = 426,000 pixels that represents 426000/15,000,000 of the total area = just 2.84% of the sensor. Even a typical point and shoot has a sensor area around 9% of that of a 40D/50D.
Why on earth would anyone expect such a tiny area of the 50D's silicon to have as little noise as the much larger area from the 40D? It just does not make sense that it would. It follows that comparing individual pixels is a recipe for disappointment where the 50D is concerned. It is not comparing apples and apples. It is comparing apples with 2/3 of an apple. What you should do is to compare the same total area of the sensor from each camera and then see which one produces the best balance betwen noise, detail, DR etc..
I hope that makes sense. It also demonstrates why I don't think you do get more "cropability" from the 50D than the 40D. What you do get, for any equivalent crop (in terms of area, not pixels), is more detail in the image.
I find it very interesting about the use of non-standard ISO's and the camera's manipulation. Do you know if this is also how the Nikons work, eg D300, D700?
I'm not au fait with stuff outside the Canon brand. From snippets I've read I think the Nikons might be OK. i.e. all ISOs are generated by analog amplification at the sensor rather than digital manipulation later on.
Canon's 1 series bodies have "proper" ISOs throughout the standard ISO range, with analog amplification to achieve them. The Canon "L" (50 ISO) setting is a fabrication, even on 1 series bodies. I'm not sure about the "H" (6400 ISO) setting.
I have got my 50D set up to only increment ISO in full stop jumps rather than 1/3 stop so that I do not waste my time going through redundant ISO settings. I do not use Auto ISO at all since that will pick whatever ISO value it likes, including the intermediate "fake" ISOs.
Tim,
I agree with what you say however I believe that for sensor performance comparison a photo has to be viewed and examined at 100% or 1:1, i.e. one pixel on sensorr = one pixel on screen, at any other size data will be interpolated and thus comparison will not be valid. This is why all the review websites compare photos from different sensors at 100%. If the goal is to end up with a 1000X800 pixel image for web posting, any current DSLR can do a good job and you wouldn't need 15 mpixels any ways, in fact my old 20D could easily do 1000X800 pixel sharp images. The purpose of 15 mpixel for birding is for cropping and being able to end up with a usable photo from farther/smaller birds without needing to log a long and havey lens otherwise there is no point in such high density sensor :eek:
There is little doubt that you can salvage good photos from 50D as shown by Doug :) but then again you can do so with the 40D and you will need less NR :D the point I was making was that 50D has little advantage over 40D in IQ department, not that it is a bad camera on its own. Also regarding AF, I guess to greatly depends on your technique and lens, I think 50D AF sensor might be calibrated better than the 40D so when using with thin DOF lenses like the 500 and 600 f/4 it gives a higher number of in focus shots but in terms of tracking capability especially locking on the BG when tracking a bird against foliage conditions it is very similar to 40D for some people like me, I also read the Canon white paper and did not see they mention any particular improvement in AI servo http://www.usa.canon.com/uploadedima...20II%20WP2.pdf In 40D whitepapre they mentioned a 30% increase in AF speed in AI-servo mode compared to 30D/20D, a claim that I found to be about right, any ways I wonder what the next model will bring. If it is going to be the same plus say a 20 mpixel sensor I will be done with XXD serie.
Quote:
Originally Posted by Tim Dodd
Quote:
Originally Posted by Sarah Jessup
In Canon cameras (XXD and 5D MKII) the intermittent ISOs are achieved by scaling of the RAW data, you can verify this by downloading this application http://www.cryptobola.com/photobola/rawnalyze.htm and then opening a RAW file while holding the CTRL key in Windows. It shows you the histogram for 14Bit RAW intensity levels (not to be confused with a RGB histogram) gaps in the histogram for 1/3 stop ISOs indicate this scaling. This leads to increased noise and artifacts in some cases, here is an example from 5DMKII at ISO 500, shadows show noise even after some chroma NR and there is also vertical banding.
http://www.stanford.edu/~ahazeghi/Ph...-143661584.jpg
Shadow area
http://www.stanford.edu/~ahazeghi/Ph...-143663584.jpg
These artifacts are not usually present at full stop ISOs. I always use only full stop ISOs with my 50D and MKII, for this reason. Nikon cameras (D300, D700/D3) use analog sensor gain amplifier to achieve 1/3 ISOs and don't show this issue.
Arash, I completely understand the argument for comparing at 100% in order to avoid interpolation artefacts but I have two points to make regarding that comparison....
1. Viewed at 100% the magnification is significant, taking the images generated far beyond photographic norms for assessing IQ. Because the magnification is so great it completely alters the DOF calculations and significantly impacts on the consequence of misfocus, lens softness, diffraction and acceptable shutter speeds to conceal camera shake and subject blur.
Assuming a monitor with a pixel pitch of 96DPI (not unreasonable), a 50D image file of 4752 x 3168 pixels, viewed at 100%, would generate a virtual image of 4752/96 x 3168/96 = 49.5" x 33". That's over 4' on the long side.
Doing the same thing with the 40D's 3888 x 2592 pixel files the virtual image would be 3888/96 x 2592/96 = 40.5" x 27".
That means you are looking at the 50D file with a magnification factor 49.5/40.5 = 1.22X greater.
Now, if you want to compare apples with apples that comparision is simply not fair, if you view both examples at the same viewing distance. You should step back 1.22X further in order to view the 50D files. Given the crazy degree of magnification for both, you should probably view the files from the 40D at a distance of around 4' and the 50D files at a distance of around 5'. Then you would have a fair comparison.
2. Moving back to the real world rather than the laboratory, isn't interpolation a reality of normal photography? If you want to judge real world image quality is it not reasonable to view the files after pushing them through real world processing? Either you will print or you will display on some sort of monitor.
If you print, surely you will print files from either camera at the same size - 12x8 for example. That's how they should be compared. Make it 24x16 if you prefer, but make the playing field level.
If you want to prepare images for a slideshow then surely somewhere in the workflow the files will be resized to fit the display device. Let's say you will display uncropped images on a 1920x1080p HDTV. Both files will need to be interpolated. That is real world use. What is not real world is downsizing the 50D files to match the 40D, nor is upsizing the 40D files to match the 50D. I personally do not agree with upsizing. I know some people do it but you can't invent detail that is not there. That approach really does put you at the mercy of the software. So why not downsize both files to some common and practical size?
If you target the files for display on a 1920x1080 device then the 50D files will be downsized to 1080/3168*100% = 34% and the 40D files will be downsized to 1080/2592*100% = 42%. That seems fair to me. Both files get interpolated. The playing field is level.
Staring at 100% crops from both cameras at the same viewing distance is, in my opinion, not a valid comparison. Here are 100% crops from my 1D3, 40D and 50D of the same subject, taken with the same 85mm lens, from a tripod at the same subject distance, each at 800 ISO....
1D3
http://lh4.ggpht.com/_4_R8TkwT74w/Si...op100_0004.JPG
40D
http://lh5.ggpht.com/_4_R8TkwT74w/Si...op100_0004.JPG
50D
http://lh5.ggpht.com/_4_R8TkwT74w/Si...op100_0004.JPG
For the life of me I cannot understand how anybody could claim that the 50D image is not magnified more than the 40D image when viewing both at 100%.
Tim, but at 100% there is no "magnification" one pixel on sensor corresponds to one pixel on the screen, regardless of what the size of image is, be it 100X100 pixel or 100,000X100,000 pixels, anything other than 100% is "magnification" greater or smaller than unity, also I would not buy a higher megapixel camera to print at the same size:o, when I pay for extra megapixels (in the case of my 5DMKII) I do this to utilize all pixels and make a larger print otherwise it would be just wasting electrons and money :D Any current DSLR on the market from the $700 Rebel to the $7000 D3X can produce similar and excellent images at sub 3000 pixel resolution, a 15 mpixel camera should have 15 useful megapixels not that you have to downsample the files to hide the imperfections.
Any ways, the reason I bought 50D is for birding and being able to do heavy crops for small/far birds but so far I have failed to see it giving me the "reach" that I expected, it maybe that my modest 400 f/5.6 is not as good as the 500 and 600s people here are using but I am going to keep it for now and try to improve my skills :)
Quote:
Originally Posted by Tim Dodd
Arash and Tim, I am no scientist or engineer - just a retired mouthpiece about to buy the 50D for BIF.
I live in a caravan; I will never print anything larger than A3 - most of the time A4.Quote:
Any ways, the reason I bought 50D is for birding and being able to do heavy crops for small/far birds
Is it incorrect thinking that some of the reasons to purchase cameras with more megapixels is 1) it will permit substantially heavier crops and still result in a higher IQ than a smaller megapixel camera, and 2) in a wide angle landscape image that is not heavily cropped, the resulting print from the higher megapixel will - everything else being equal - present a higher IQ than a smaller megapixel camera?
If I shoot the same scene with the same settings et cetera and I do two images, one with a 10 megapixel camera and one with a 20 megapixel camera, will the 20 megapixel camera present a higher IQ?
Arash, regarding reach, and again I am asking questions to learn and might say it incorrectly, with the 400 5.6 you are limited to 640 because you cannot add an extender; I am going to use the 50D with a 300 + 2.0 extender and that should give me 960, correct?
Am I correct that you lose AF when you exceed f/5.6; that is what I was referring to in the preceding sentence about being limited to 640 with the 400 f/5.6 and the 50D.
I ask because on the one hand I am considering the 500 f/4 which would be limited to the 1.4 = 700 + crop factor = 1120 and I am wondering if the increase from 960 to 1120 (assuming my math is correct) justifies the added expense, physical length and weight to spend an additional $1,700 over the 300 f/2.8 for the 500 f/4.
It is definitely not something I am going to purchase today; is the IQ from the 500 + 1.4 that superior to the 300 +2.0?
I hope I made sense!
Cheers,
Thx for all the Help here appreciate it lots. Im really starting to think that i have a bad copy of the 50D. will do some test in the next day or 2. Thx all
Arash, the sensor has a physical size. For a 50D it is 22.3mm x 14.9mm. That's the physical size of the chip that captures the light. Pixels have got nothing to do with it at this point. However, when you view at 100% you are generating a virtual image which, in my example earlier, is 49.5" wide. You do not get from 22.3mm to 49.5" without magnifying something. 22.3mm = 0.88". The magnification factor is 49.5/0.88 = 56X. That is a colossal magnification.
Doing the same maths for a 40D the magnification factor is 40.5/(22.2/25.4) = 46X. That's still a lot but nowhere near as much.
If you now think about the good old days of 35mm film you had a frame of 36mmx24mm. That might typically be printed at a size around 10x8" to check quality. 36mm = 1.42". The magnification factor to go from a full frame negative to a 10x8 print is 10/1.42 = 7X.
So in the old days people might shoot at up to 400 ISO, typically, and magnify their images no more than 7X to view the quality. Today people are magnifying their images 56X, shooting at 1600 ISO and complaining about noise. Sheesh!
p.s. If you want to look at this another way, you agree that pixels on a 50D are smaller than those on a 40D - true? So when you map those pixels to physical pixels on a display monitor, tell me, are you or are you not magnifying the 50D's pixels by more than the 40D's? If you do the measurements at the micron level I think you'll find you are magnifying the 50D pixels more.
I think the bottom line in all this, for simplicity's sake, is that image quality comes from sensor area. The bigger the sensor the more light captured. 5D2>1D3>50D. The more light captured the higher the image quality. When you start cropping all you are doing is discarding some of valuable sensor real estate you paid for. A crop in software is really no different to starting out with a smaller sensor in the first place. If you crop tighter from a 50D file than you would from a 40D file you will end up with the results from a smaller area of the sensor. Your IQ will reduce. By pixel peeeping at 100% all you will do is magnify the problem you brought upon yourself by cropping more.
I too have had frustrations with noise and lack of cropability with my 50D. After this and other great threads on BPN I am beginning to understand why, what I can do about it, and what it is realistic to expect.
I greatly appreciate Tim Todd's logical and thoughtful explanations. I have a question that emerges from that, regarding monitor size. It would sound as though going for a large monitor may not be productive after a point because the image will be enlarged too greatly, especially when working in Photoshop and viewing slide shows.
With the 50D, that limitation may be greater. Input appreciated. Thanks
Very erudite discussion. I'm learning a lot myself.
Glen, my usual monitor for everything I do, including picture editing, is a 1920x1200 17" laptop display. That suits me perfectly and I love having the high resolution to aid editing, viewing and having toolbars and tool palettes open and still being able to see enough of the image I am working on. Even with files from my 40D and 1D3 I have to view them at no more than 50% - a little bit less actually, in order to fit the whole image onto the screen. With files from my 50D I can almost fill the screen when viewing at just 33%. I do not think there would be a problem in going up to a significantly larger display, so long as the monitor has enough resolution to not look too pixelated.
I do not print my images, ever. For use at home I display them on my 40" 1920x1080 TV. The picture is huge. The photos look fantastic. But I view from about 8'-9' away so the picture detail pretty much outresolves the acuity of my ageing eyes. In other words, a 40" display is fine when viewed from a suitable distance. However, I wold not want to work in front of a 40" display as my desktop monitor. I'd get neck ache scanning the thing and would not be able to see the whole image at once without leaving my seat and moving several feet back.
The point is, get whatever monitor size you feel comfortable with and a resolution that satisfies your eyesight without strain. I certainly would not let fears about the IQ from cropped or uncropped photos stand in the way of getting what I wanted. How big are you talking - 24"? 30"? When viewing slideshows would you expect to sit 18" from the screen or settle back in your chair and view from 3' away or more? Consider that a 24" screen viewed from 3' away will look no bigger than a 12" screen viewed from 18".
Let me put it another way. I use my laptop on the coffee table in my living room. Right now the TV is about 8' away from me, the laptop screen is roughly 18" from my eyes. In relation to my field of view the 40" TV is absolutely dwarfed by the 17" laptop screen. In terms of visible screen width the TV screen looks to be only about 60% as wide as the laptop screen, yet the screen is actually well over twice as wide.
Size isn't everything. It's how you use it :)
Thanks Tim,
I have a 19 inch and was thinking of a 22 or 23. Obviously highest resolution possible is important. I also will think about a top quality laptop for photo editing purposes. Glen
I am personally not convinced that that is totally accurate. I believe the quality comes from the amount of light falling on EACH sensor PIXEL, so the bigger area the pixel gets its light from the better, not the bigger the sensor. The 5D2, 1D3, 50D would have the same image quality *IF* they had the same pixel density (which they do not). Also the more pixels the closer a pixel is to another, and in general terms, the more noise.Quote:
Originally Posted by Tim Dodd
Tell
Tell, you are falling into the trap of comparing PIXEL quality instead of IMAGE quality. By your logic, no matter what size the sensor, so long as the pixels were the same size then the quality would be the same. The quality of the pixels would be the same, yes. The quality of the images would not.
Here's an example..... Take a 5D2 and shoot a composition to fill the frame, say at 400mm. Now stay where you are, half your focal length to 200mm and shoot again. Now half your focal length once more to 100mm and shoot again.
- Take the image from the first full frame shot and print those 21 lovely megapixels as a 30x20.
- Take the second image and crop it to 1/4 of the frame, to maintain the exact same composition. You'll have approx 5 equally lovely megapixels remaining. Then print the crop as a 30x20.
- Take the third image and crop it to 1/16 of the frame, again to maintain the exact same composition. You'll now only have just over 1MP of beautiful, noise free pixels remaining. Then print the crop as a 30x20.
According to your logic the quality of each print will be the same. According to my logic the quality of each print will be nothing like the same. What's the difference - not pixel size, density or quality; that's identical. The difference is in sensor size.
Again, if you stop thinking about pixels and concentrate your attention on images, IQ comes primarily from the size of your sensor real estate, not the size of your pixels. Remind me, it is the images we are interested in at the end of the day, is it not?
Why is it that the IQ is little different between Rebels and xxD bodies, and not vastly different from one generation of body to the next? Why is it that the IQ from the 1D3 is really wonderful compared to the little croppers? Why is it that the 5D hammered the croppers for IQ and the 5D2 has certainly not taken a step back, despite the massive increase in pixel density and reduction in pixel size?
The primary driver for IQ is the size of the sensor. Relatively speaking it matters little how many pixels you stuff onto the thing. The big leaps in IQ come from big leaps in sensor size, not from fiddling around with pixel size/density.
Of course, a bigger sensor needs longer glass, and although IQ goes, up the costs, bulk and weight of the systems mount even faster. So, if the size of the sensor has no bearing on IQ, why do people even bother buying into the more costly, more bulky, heavier, larger format systems?
p.s. The 5D2 and 1Ds3 have the same pixel size/density as the 20D and 30D. Would you say that the IQ from the four cameras is equivalent? Can a 30x20 print from a 30D match a 30x20 print from a 1Ds3?
Hi Tim, ok we are talking from slightly different perspectives, I will maintain what I said because I am talking about image/pixel quality as seen at the SAME resolution (at 100% as I think the op was referring to), by your printing method you are adding another dimension which distorts the result IMHO, perhaps the difference between us is that I am looking at image quality as seen on the computer screen and you are looking at quality as seen in a print ?
Tell
Tell, my interest is in whether or not I can achieve the end product I want, to an acceptable quality. While the quality of individual pixels might hold some academic interest, for me that's not really the point. The point is, can I get the image I desire, at the size I wish, on the medium of my choice, and like the way it looks. If the end product is a print or a webshot or even a slideshow image then I don't really feel that staring at individual pixels is helpful or necessary.
Rant aimed at nobody in particular follows....
Compared to the good old days of 35mm film the crop body cameras like the 50D are already significant crops. They have sensors with only ~40% of the area of a frame of 35mm film. That's a big crop to start with. Then people want to crop into that, not a little, but a lot, perhaps throwing away 75% of the pixels or maybe more. And still they don't want to see IQ diminish. Well, sorry, guys. Right now we're still on planet earth and bound by traditional laws of physics. You can't really expect high ISO, low noise and high detail when you are working from an area of the sensor equivalent to just 10% of a standard 35mm frame.
Stop blaming the camera because you feel you have to crop too much. Look at your glass. Why not blame that for being too short? Or why not blame yourself for not getting closer to your subject? Why does the camera get the blame?
If you really want IQ to increase, stop cropping. Go get some longer glass or find a way to get closer to your subject. I see no reason that you should expect IQ from the 50D to equal that of the 40D by cropping even more aggressively.
I've given up blaming my equipment. I've also pretty much given up shooting BIF. Why? Because I do not have sufficient interest or dedication to get up at ridiculous hours of the day, or sit in a hide for hours on end, or learn the habits of my targets etc. etc.. As far as birds are concerned I am an opportunist photographer. I take my camera out when walking the dog. Walking dogs and approaching close to birds are not very complementary pursuits. As a consequence I never get near enough to my prey. I used to shoot with my 100-400 and 50D but always the birds were too small in the frame. I would try cropping but I was never really happy with the result. Is that the camera's fault? Really it isn't. The fault is with me for being lazy, not putting in the effort and just expecting opportunity to be handed to me on a plate. Not gonna happen.
There are other issues that concern me regarding the 50D's pixel density and people's expectations for IQ, specifially related to photographing things that move, but right now I am tired and I suspect you guys have heard quite enough from me, so I shall bid you goodnight.
Tim, personally I have not "heard quite enough from" you; thank you for all of the time you have taken in this discussion. You have given us a lot of food for thought; especially with the next comment. Cheers Mate.Quote:
There are other issues that concern me regarding the 50D's pixel density and people's expectations for IQ, specifially related to photographing things that move, but right now I am tired and I suspect you guys have heard quite enough from me, so I shall bid you goodnight.
Tim, the description "opportunist photographer" is a wonderful way of thinking about yourself in relation to photography in general. It is the same but different approach that was discussed in a thread by James Shadle pertaining to what it takes to be a great photographer. (I can't find that thread - anybody?). When I read and participated in James' thread at the beginning of my membership on BPN - just a few short months ago - I really had to think hard about what I wanted out of my newly chosen photography experience, and it is an evolving pursuit.Quote:
I've given up blaming my equipment. I've also pretty much given up shooting BIF. Why? Because I do not have sufficient interest or dedication to get up at ridiculous hours of the day, or sit in a hide for hours on end, or learn the habits of my targets etc. etc.. As far as birds are concerned I am an opportunist photographer. I take my camera out when walking the dog. Walking dogs and approaching close to birds are not very complementary pursuits. As a consequence I never get near enough to my prey. I used to shoot with my 100-400 and 50D but always the birds were too small in the frame. I would try cropping but I was never really happy with the result. Is that the camera's fault? Really it isn't. The fault is with me for being lazy, not putting in the effort and just expecting opportunity to be handed to me on a plate. Not gonna happen.
I think you are spot on regarding "stop blaming the camera" and look within.
To learn to chase the elusive BIF I have moved from the 40D/100-400 with a maximum AF focal length of 640 to the 300 and I am about to acquire the 50D which will give me a maximum AF focal length of 960. That dividing line between those willing to be sufficiently/significantly dedicated enough to be great photographers and those happy to be opportunist photographers is perhaps what this thread - in a nontechnical sense - is all about.
You rightly raise the question in response to a cry "Please help Canon 50D" whether it is the 50D that needs the help. (Kobus, my response has nothing to do with you personally - perhaps you do have a faulty camera.)
Artie obtains amazing images with his 50D and his 400 DO lens; he also presents an amazing number of images with his 1D3 + 800 f/5.6 + 1.4.
When we have Artie's years and dedication then and only then will our XXD and 1DX images come close to Artie's images. Artie, Doug, and other Master Professionals know and know how to work within the limitations of their cameras.
Until we, the students, have their level of experience, I think you are right:
The corollary to your statement is to learn your camera and learn how to push it to its maximum capabilities without going beyond and then blaming the camera for not keeping up with our pushing.Quote:
Stop blaming the camera because you feel you have to crop too much. Look at your glass. Why not blame that for being too short? Or why not blame yourself for not getting closer to your subject? Why does the camera get the blame? If you really want IQ to increase, stop cropping. Go get some longer glass or find a way to get closer to your subject.
And in my case, this wonderful thread has caused me to question whether I will be purchasing anything longer and heavier than my 300. Perhaps, when I too can manually focus quickly to justify a 500 f/4 + 2.0 or an 800 f/5.6 + 1.4..........just perhaps :D
Thanks again, please keep the comments flowing - greatly appreciated.
Cheers, Jay
Tim,
Unfortunately I don't have the time to read all of the extensive followups in here but for scientific purposes sensor performance should and is evaluated at pixel level, anything else is just a waste of time. when we measure Quantum Efficiency of the sensor in the lab we look at individual pixels to measure internal and external quantum efficiencies, the total area of the sensor does not matter at all, for example if you have a large sensor but with small pixels versus a small sensor with large pixels the image coming from the small sensor with large pixels will have lower per-pixel noise and will look cleaner albeit it will have lower resolution. It all depends on pixel area not sensor area, you can take 25 small sensors from a regular digicam and stitch them together to make a huge sensor but IQ will not be any better :)
One pixel is one pixel and when you do measurements one pixel on sensor should be one pixel on the screen. By the laws of quantum mechanics, everything being equal a smaller pixel will always have higher shot noise than the larger pixel thus a 50D image always has higher noise than a 40D image at 1:1 size, there is no magnification whatsoever.
P.S. If you have access to technical journals and want to find out how we measure QE see this excellent articale published in several IEEE journals, unfortunately I cannot post any of the content due to copyright issues.
A Method for Estimating Quantum Efficiency for CMOS ImageInformation Systems Laboratory, Stanford University
Sensors
Boyd Fowler, Abbas El Gamal, David Yang, and Hui Tian
Jay, this is why I buy higher megapixel cameras too, to be able to crop more or make larger prints. Thats is why I got a 5DMKII, otherwise I already had an excellent 12 mpixel camera, the D700.Quote:
Originally Posted by Jay Gould
This is when it gets tricky, more megapixels do not necessarily mean higher IQ, as much as the number of pixels matter, the size of the pixels matter too. Imagine camera A has 10 mega large pixels and camera B has 20 mega small pixels, then camera B's pixels have a noisier signal and there will be a trade off, you get more noisy bits versus less clean bits. In this case you need to compare pixel size between the two that is to devide the sensor area by # of pixels and take the square root to get the pixel pitch. If the pixel pitch is equal or less than about 4.5um with current sensor technology and optical lenses, you are asking for trouble. 50D has a pixel pitch of 4.7um which is very close to this limit, still much larger than common digicams that have a pitch of 1-1.5um. a Canon 40D has a pitch of 5.7um, a 5DMKII has a pitch of 6.4um and a D700 has a HUGE pitch of 8.45um. so between 40D and 50D there is little gain because you are hitting the limit, but between 5DMKII and D700 you do gain extra resolution because you are still in the safe territory, although you lose a bit in DR and noise performance but the increase in resolution is obvious and worth the drawbacks... As sensor technology matures we see the minimum pixel pitch further scale down and we will be eventually limited by the performance of the optics and diffraction, 50D is already close to this limit.Quote:
If I shoot the same scene with the same settings et cetera and I do two images, one with a 10 megapixel camera and one with a 20 megapixel camera, will the 20 megapixel camera present a higher IQ?
Quote:
Arash, regarding reach, and again I am asking questions to learn and might say it incorrectly, with the 400 5.6 you are limited to 640 because you cannot add an extender; I am going to use the 50D with a 300 + 2.0 extender and that should give me 960, correct??
correct but you have to ask people who have the 300 + 2XTC if the AF and optical performance is acceptable, to me 2X TC is too much, especially for 50D.
yesQuote:
Am I correct that you lose AF when you exceed f/5.6; that is what I was referring to in the preceding sentence about being limited to 640 with the 400 f/5.6 and the 50D.
I guess 500 with 1.4 is a better choice because 1.4 X TC will not slow down the AF or degrade IQ as much as the 2X does :)Quote:
I ask because on the one hand I am considering the 500 f/4 which would be limited to the 1.4 = 700 + crop factor = 1120 and I am wondering if the increase from 960 to 1120 (assuming my math is correct) justifies the added expense, physical length and weight to spend an additional $1,700 over the 300 f/2.8 for the 500 f/4.
It is definitely not something I am going to purchase today; is the IQ from the 500 + 1.4 that superior to the 300 +2.0?
I hope I made sense!
Cheers,
On that basis, would it be fair to say that people who stitch together frames to form a panorama, or to produce high resolution images of the moon are also wasting their time? If so, I wonder why on earth (or the moon) they bother. Perhaps you should point out to them their foolish ways.Quote:
Originally Posted by arash_hazeghi
Take, for example, this image, prodced by stitching together 15 frames from a 20D. Click the image for a larger file....
http://www.dslreports.com/forum/remark,16822681
See the first post on this page for a description of how the image was created. Also note that before downsizing the combined image reached 40MP (from an 8.2MP camera)....
http://www.dslreports.com/forum/rema...22681~start=53
I guess that won't print any larger than 12x8 then :(
Tim, you are confusing noise and resolution, a panaroma is to produce a large file with more detail it is not about reducing noise or increasing sensor QE. I stated above that sensor QE performance is measured per pixel basis, you are saying that people create panaromas to produce a larger file, I see no causal relationship between the two statements. Any ways I provided a technical article which sheds more light on this issue if you are interested just have a look...
good luck with your gear :)
OK, after the cliffhanger ending last night this will probably be a huge anti-climax, but anyway.....Quote:
Originally Posted by Jay Gould
While we can all argue about per pixel quality and per pixel noise I hope we can all agree that given the right circumstances the 50D sensor is capable of producing very high resolution, very detailed images. At 100 ISO, on a solid tripod, MLU, remote release, static subject, no wind, perfect focus, sharp lens, modest focal length, stopped down a bit, but not too much, plenty of light, good contrast, clear air, etc. etc., every one of those little pixels can be made to count. That is pretty much the ideal, for any camera. But due to the high pixel density of the 50D these things become even more critical in order to yield the maximum IQ attainable.
The very thing that is the 50D's blessing - the resolution - is also its curse, because if anything at all is even a little bit off then, of all Canon's cameras, the 50D is the one that will highlight the error first. If you get a little bit of camera shake the 50D will make it show, more than the 40D and far more than the 1D3 or 5D2. If your subject moves, even a tinsy winsy bit then the 50D will be the first to show the blur, when you stare at the pixels at 100%. If your focus is off, just a hair, then the 50D will expose the flaw first. If your glass is soft, or you stop down too much (into diffraction softening territory), the softness will show on the 50D more than any other camera - when you view the image at 100%. There is nothing at all wrong with the camera. It is doing its job perfectly. But if there is a flaw anywhere else in the image gathering chain then the 50D will expose it first.
So, what on earth is the point on the 50D? Well, remember, not everyone shoots birds and not everyone shoots BIF. Some people shoot less demanding subjects, where they have time and control and can tighten things down or work with a huge DOF. Not everyone crops the living Bejesus out of their images. Think landscapes, architecture, product photography, portraiture, automotive (not motorsports), macro, anything in a studio. In these disciplines you can use a tripod. You can use 100 ISO. You don't need a crazy high shutter speed. You can take an age to fine tune focus. You can (often) supply your own light, probably flash/strobe. You can shoot with a modestly priced prime lens. You don't need to crop. In these conditions the 50D is easily the cheapest way to obtain the most detail in an image, the image with the largest print potential etc.. Remember, Canon did not design the 50D for the sole benefit of birders. It is a general purpose camera that must meet the needs of an exceptionally wide audience. Even in motorsports, or field sports, or showjumping you have (relatively speaking) great big hulking subjects that move relatively predictably, and in scale terms quite slowly, and you probably do not need to squeeze out the finest sub-millimeter details from a race car.
Now, what happens when you put the 50D in the hands of a birder? Well, all of a sudden the game is changed dramatically. You have oncooperative, flighty subjects, forever wriggling and moving about. They pick the oddest of places to perch, with miserable light. They're normally far too small and too far away and you need a massive lens, which costs a fortune and brings about its own challenges for utilisation. And so far we're still only looking at birds on a perch or wandering about. Now consider the buggers when they take to the air. Their bodies are moving in one direction. Their wings are moving in another. You can't pan with all the movements at once. You need huge shutter speeds. Long glass isn't that fast so you need to bump the ISO. All of a sudden there are demands placed on AF performance, long lens technique, panning ability, DOF, ISO/noise that many photographers simply do not face. What might look good at "normal" viewing sizes is more than likely going to look less than perfect when viewing pixels at 100%. Who cares about per pixel noise when you have blur or softness in the image that spans two or three or four pixels?
So, if you compare a BIF image from a full 50D frame with a BIF image from a 40D full frame I expect the results will look very comparable. If the 50D does have an edge with the AF performance, and if the AF microadjustment enhances the focus accuracy too, then there is a chance the 50D image may even look a little bit better. But, if you start cropping away, espcially to the point where you are looking at individual pixels, the 50D will almost certainly look softer and noisier than the pixel level view from the 40D. The resolution advantage the 50D gives you in controlled conditions becomes its undoing when the conditions are less controlled, and yet some people still insist on magnifying the 50D image by 22% more compared to the 40D.
If you really want to compare at pixel level then you'll need 22% less shake in your camera/lens. You'll need 22% less movement in your subject. Easy fix - increase shutter speed by 22% (basically 1/3 stop). But now you'll need to bump your ISO a notch, which increases noise, and that's not really fair because we have already magnified the image more, so really we should be knocking back the ISO by 1/3 stop, which means you need to open the aperture by 1/3 stop. But you may be wide open already, and if not then your lens IQ may drop is you open up more. and so on and so on.
So, with all that said, I'm not sure the 50D does offer much, if any, realisable advantage in cropability over the 40D, for BIF. I just doubt that people can easily and repeatedly capture a pin sharp, pixel perfect, blur free, low pixel noise image of a BIF with a 50D. Maybe those with amazing skills/talent and a little luck can up the percentages, but if you're anything like me, the chances are slim. Of course, there will be times when all the planets are aligned and you pull off a stunner, but not each and every time you shoot a frame.
To go on just a little further, and this is not specific to the 50D, but rather to cropping in general, consider the rule of thumb for hand held shooting and shutter speed. For a 35mm frame it's nice and easy....
Shutter speed >= 1/focal length
e.g. to handhold a 400mm lens your shutter speed should be 1/400 or faster. Some people can manage half that, others need it faster still. I'm guessing here but I suspect that people have improved stability when standing calm and relaxed and pausing their breathing, while aiming at a static subject. When it comes to panning across the sky I am not sure whether the challenge gets easier or harder. I somehow doubt it is easier, but I do not know. What is for sure is that, with a moving subject, not only will your camera/lens technique need to be up to scratch but so will your panniing skills. Pan too quickly = subject blur. Pan too slowly = subject blur.
For a crop camera the rule is modified to take account of the crop....
Shutter speed >= 1/(focal length x crop factor)
So to use a 400mm lens with a 40D or 50D you should be thinking of using a shutter speed of 1/640 or faster. OK, that seems fair enough, but what if you then perform a software crop on your image file? Well, you've just changed the crop factor again. If you crop to 1/4 of the image you have doubled your crop factor. Now you'll (probably) need a shutter speed of 1/1250 or faster.
OK, that's all well and good. I'm not sure what size of enlargement that old rule was designed for - let's assume a 12x8 print from a 35mm film frame. Now let's check out the 50D again, assuming we've cropped to 1/4 of the frame. Now we will have an image of 2376x1584 pixels. Let's assume our monitor has a pixel density of 96PPI. Viewed at 100% that cropped image will be 2376/96 = 24" wide. Hmmmmm! That's pretty much exactly twice as big as that old 35mm enlargement. That's probably going to expose camera shake if the shutter speed is not fast enough, probably subject/panning blur too. It seems like we need to double the shutter speed again in order to stand a hope of sharp images at the pixel level, and that's assuming the AF was faultless too.
So, our baseline shutter speed for sharp BIF images viewed at 100% is 1/2500. At best our lens only opens to f/5.6. Assuming bright, direct sunlight and using the Sunny 16 rule as an exposure benchmark, to achieve 1/2500 at f/5.6 we would need an ISO of 320. Now, that sounds pretty good. In fact, why not keep things neat and tidy and go for 400 ISO and 1/3200? That should hold detail in white plumage in bright, sunny conditions. But what if the light is not bright sunshine? What if it is early morning, or late evening, or a little bit hazy, or cloudy, and the sun is not at full strength. Maybe it is a stop down from "Sunny 16", maybe a smidge more. So now we are looking at 800 ISO at a realistic shooting speed. But Uh-oh! Noise! What if our subject/scene has no bright highlights? What if it is a neutral scene of a brownish or black bird against green foliage? Should we expose to the right a little? If so, perhaps we should add another stop to the exposure. So now we're at 1600 ISO. Ouch! But wait, f/5.6 does not give enough DOF, so we need to stop down further, and bump the ISO some more. We're viewing at 100% and the noise is going to show up bad. What can we do? I guess there are some choices here....
- complain that the 50D is crap and not fit for the job;
- try running some NR software;
- stop viewing at 100%. Instead, fill the frame with your subject more, view at 50%, or whatever your final image size should be, run the NR software anyway, and stop fussing over the wretched pixels;
- stop trying to take shots in "impossible" circumstances and blaming the camera when your results are disappointing.
One last point, which I'll try not to labour - AF with a lens slower than f/2.8 only has a specified accuracy of "within the DOF". Even with AF microadjustment there is no certainty that the AF will plonk you in the right spot time after time. Compound that technical reality with the demands of AF tracking, both by the camera and the photographer, and the chances are that, viewed at 100%, the AF will be a little off. Remember that Canon definition of "accuracy within DOF" is based on the DOF calculations of a normal sized print (12x8?) viewed from 10-12" or something like that. Apparently it's all in their Lenswork document. When you view at 100% the whole DOF thing is changed completely. What might have looked sharp as a 12x8 may very well not when viewed at 100% as a window into a virtual image roughly 4X larger (16X by area).
I hope that some of my ramblings make at least a little sense.
Happy birding :)
p.s. the way I apply these thoughts to my own birding efforts is that for perched birds I'll use my 50D, and probably a tripod, to squeeze out the last drop of detail I can from my inadequate lens and inadequate fieldcraft. For BIF, quite honestly, I find my 1D3 to be the better bet. Not only is the AF superior, in my opinion, but I don't have gazillions of worthless pixels wasting their time recording a load of pointless blur. Stick on a 1.4X converter and the centre point AF is still very sweet. I benefit from genuine image magnification with the extra glass, yet avoid the pixel level noise and pixel level blur of the 50D. I also get a far nicer viewfinder, more controls to fiddle with, freedom to use intermediate ISOs, higher framerate and a nicer experience all round.
p.p.s. My birding lens is my 100-400. That puts extra constraints on me. Wide open at 400mm (some say it's actually only around 380mm at full stretch) it's not going to compete with the 400/5.6 prime, never mind the 500L or 600L, either for sharpness or AF performance. If I pop on the old 1.4X teleconverter for the 50D then (a) I lose AF; (b) even wide open I'm at f/8; (c) if I stop down even just 1/3 stop to improve lens sharpness I am pushing myself further into diffraction softening territory. Apparently, according to the scientists, the 50D hits diffraction softening limits at f/7.x, so for birding it's pretty much futile to stop down below f/8. Talk about a rock and a hard place.
p.p.p.s Let's suppose you have shot an image which, viewed at 100%, shows a blur/softness equal to 1.5 pixels in width. At 100% the edges will not look sharp. If, however, your subject is so large in the frame that you can comfortably resize the image to 50% and still have a subject that is amply big enough that edge blur will largely vanish, certainly becoming far less objectionable. Now suppose your subject was so big in the frame that you could resize to 25% and still get a nice composition. Blur gone. Image sharp. Job done. Don't crop. Don't get longer glass. Get closer to your subject. All your problems will disappear, including the noise, and it won't cost you a penny.
p.p.p.p.s None of the above makes the 50D a worse camera than the 40D. When used correctly and not taking liberties with the cropping, it should be at least the equal of the 40D in terms of IMAGE quality. Enhancements like AF microadjustment, higher resolution screen, higher resolution preview image, better Live View, perhaps slightly better AF, should all swing things in favour of the 50D. Inappropriate cropping will not.
p.p.p.p.p.s. The examples I used above to illustrate the need for high shutter speeds were a kind of worse case scenario, based on hand holding an unstabilised 400mm lens, used by a photographer of "average" skills. Of course, in practice, many people can and do use IS or a tripod to help out, and some people are undoubtedly better than average at their craft, so things should improve. Nevertheless, I still believe that viewing a 50D file at 100% does require 22% less camera shake, 22% less subject blur, 22% greater AF accuracy and 50% less noise per pixel in order to look just as good as an equivalent file from a 40D. I hope my maths and reasoning have not let me down.
Tim,
Thank you for a great reminder that bird photography is very challenging, and where all those challenges are. Thank you for your frankness ..if we were all that honest with ourselves, we would moan less and work harder at our craft. When I put this all in perspective, I find it amazing that anybody can produce the quality of images that we see regularly here on BPN. That said, I have noted that most of those images are shot with the 1DIII or 40D.. not the 50D.
You have given us a lot to think about in this thread. Certainly, one thing that it has driven home to me is that we should evaluate each new camera keeping in mind how bird photography differs from many other forms of photography, and separate the great techo achievements (which may or not be turn out to be improvements) from what OUR demands and limitations are. I'm sure that Artie and others who have honed their craft for many years, using a variety of cameras not as sophisticated as those we take for granted today, could produce outstanding images from any camera you gave them. They don't take anything for granted, they know how hard it is to produce a great image, and they apply all their skill and experience for every shot. And, they say they still have failures, and are still learning.
Thanks too, to Kobus, for voicing his frustrations. Hopefully all of us will quit moaning, make some choices and some changes, and get on with that challenge we all love; bird photography!
Thanks to all of you for making BPN such a great place to share and learn!
Cheers, Glen. If I can just add to my post #43 by throwing in a few pictures, hopefully this will illustrate, at least a little, what I am trying to get at.
Here is an image shot with my 50D and 100-400 lens at 400mm, f/8, 1/800, 400 ISO. The bird is cruising and not hard to track. I can't be certain but I would think that I would have had IS Mode 1 engaged since I would not even regard this as a panning shot. The distance is great enough to give me plenty of DOF, especially at f/8, and really there is no struggle for the AF to function. Its a clear summer day at 10:30 in the morning - lots of light, lots of contrast, if not on the bird's shaded side then between it and the sky. F/8 should have given me good lens IQ. IS should have made 1/800 a very acceptable shutter speed. 400 ISO should not be something I should fear. The exposure is good - the white cheek feathers are peaking at 99.3% in Lightroom. On the face of it everything seems to be in place for an easy capture and a great photo. Here is the full, unedited image resized to 800x533....
http://lh6.ggpht.com/_4_R8TkwT74w/Sk...13_8654_LR.jpg
At this level of magnification the bird does appear rather small. However, I can't see noise anywhere in the image and the bird looks pretty sharp, just too small. That should be no problem, since I'm using the 50D and can crop away with ease - right?
Here's a crop displayed at 50%. i.e. a crop of 1600x1067 pixels resized to 800x533....
http://lh3.ggpht.com/_4_R8TkwT74w/Sk..._8654_LR-2.jpg
It's not quite what I'd hoped for, to be honest. I can see noise in the sky and the bird doesn't really look sharp. The noise in the sky can easily be fixed with software. I can probably tweak the bird to make it look sharper too. But before I do that, let's just see what I'm really working with here. Let's zoom in to 100% to check the image quality at the sensor....
http://lh5.ggpht.com/_4_R8TkwT74w/Sk..._8654_LR-3.jpg
Well, on the plus side the bird is now a decent size, for web viewing, but not much good for printing, unless I am happy with a 3x2 print. The noise is even more obvious, but again not a challenge for proper NR software, at least in the sky. But look at the bird. It is really very soft indeed. The noise that is so obvious in the sky must also be present in the bird itself, and since the bird is mostly darker the noise will be stronger relative to the image detail. Although the bird is already very soft, if we apply NR it is only going to lose even more detail, false or otherwise. The fabled cropability seems not to be there, at least in this image.
Now, what I can do is to try to simulate how this image might have looked, captured by a 40D and viewed at 100%. To do that all I need to do is to crop 22% larger than 800x533 and then resize that back doen to 800x533. It's not an ideal example but the best I can think of. Here is the mocked up equivalent from the 40D....
http://lh5.ggpht.com/_4_R8TkwT74w/Sk..._8654_LR-4.jpg
So what's really changed? Clearly the bird is a bit smaller - 22% smaller - but does the noise in the sky look a little different? Does the bird look a little sharper? I think perhaps the answer is "Yes". At the end of the day is there much improvement one way or the other? In my opinion, any advantage the 50D might have brought us has been eliminated because the image captured was simply not good enough. It was not sharp enough. It was not big enough. Given the small size of the subject the noise, even if not obvious in the bird, is still there, and masking detail that we would wish to see. So not only is the subject too small but really I have shot with very poor lighting. It's quite hard to get the underside of a bird lit at 10:30 in the morning just two days after midsummer's day.
Blame my skills at handling the camera if you will, but really I think the problem here stems not from poor camera handling, or set up, or from using the 50D instead of the 40D. The problem is that the subject is woefully small in the frame. I have not covered enough of the sensor with my subject. Arguing the toss about 22% differences in pixel size, or pixel noise, is neither here nor there when the fundamental problem is that I did not fill the frame sufficiently with my subject. If I had been three or four times closer then how very differently this image might have turned out.
Having glass three to four times longer would absolutely not have been the answer and I doubt even a better beamer (I don't have one) would have helped me out at f/8. Indeed, I suspect some of the problem here is camera shake, but it is also very possible that on a hot day there was too much wobbly atmosphere between me and the subject. That is why I doubt that longer glass would have helped. However, a nice 500/4 would have let me double the shutter speed and halve the ISO, as well as magnifying the bird, so on balance that would have almost certainly got me a better shot.
Of course, the lighting is an issue too. This was one of my "opportunist" shots. The lighting was what it was. I made do with what nature provided. Unfortunately the bird did not bank and turn in a way that caught the light for me.
Anyway, before giving up bird photography for good, let's just see what a little basic editing can accomplish....
http://lh6.ggpht.com/_4_R8TkwT74w/Sk...54_Neat_LR.jpg
It's not great, but maybe it's not so terrible either. It'll never print larger than 6x4 but it might just be worth hanging on to, at least until I can shoot something better.
Tim, thank you both for your humor - some of it was so well written in a nontechnical sense that I read it to my wife - and for your detailed technical and nontechnical considerations pertaining to the 50D.
You have certainly given us - especially me - a lot to think about when purchasing a camera that we intend to use specifically for BIF as compared to horses and buffalos ;) .
And, when Artie presents images based upon decades of 110% dedication and skill - my hat is off to you Artie! - and says that in his hands the 50D is the best BIF camera he has ever used, we cannot assume that in our photographically newer hands it too will be the best BIF camera and that we too will produce tack sharp images with a respectful keeper rate.
I will be interested to hear some of the reactions, if provided, from the senior members of the BPN using - at least some of the time - the 50D to shoot
:eek: :eek:Quote:
the buggers when they take to the air. Their bodies are moving in one direction. Their wings are moving in another.
Of course, in all fairness to the 50D, there are lots of photographers here on BPN using the 50D as their primary camera. I am hoping to hear from them too!
Sleep time for me!
Hasta Manana..................
The 50D is great for flight Jay. Image quality issues aside, it has the best AF for flight of any Canon body IMO. As Tim points out, you need to get the bird fairly large in the frame for good image quality with the 50D (and with most other cameras for that matter). That's why I'm not a big fan of spending a lot of money on a 300mm lens for birds. In most situations, you are either forced to use a 2x or must crop significantly.
Jay, I also hope we hear some thoughts from other 50D togs, whether satisfied or dissatisfied. If I can sum up my thoughts in as few words as possible I guess it would go like this....
Compared to the 40D, the 50D is capable of delivering superior image detail and potentially increased sharpness, when used in concert with good photographic technique, good fieldcraft, good glass, good light and skilled post production, allowing larger prints to be made, or tighter crops. If any of those things are sub-optimal its advantages over the 40D may be reduced, perhaps to nil. All things being equal, it is unlikely to produce an image that is materially, if at all, inferior. With any modern DSLR camera system, the biggest influence on the success or otherwise of the results achieved is likely to be down to the person operating the gear, not the gear itself.
I hope that helps :)
p.s. Here's a 100% crop with a bit of editing. The editing is crude so there is some haloing, but it looks like the image has come along quite well compared to the raw original....
http://lh6.ggpht.com/_4_R8TkwT74w/Sk..._Neat_LR-2.jpg
Here's an easier (in terms of cropability) example, hopefully demonstrating the advantages when you can lock down movement and have the light in your favour. This was shot with my 50D and 100-400, this time with my Kenko 1.4X teleconverter fitted, and mounted on a tripod. The teleconverter pins were taped so the EXIF does not show its presence. The correct EXIF is 560mm, f/10, 1/400, 400 ISO. I used Live View Live AF (contrast detection), which is slow but deadly accurate.
Full frame without edits....
http://lh6.ggpht.com/_4_R8TkwT74w/Sk...45_1795_LR.jpg
60% crop with NR and sharpening. No other edits....
http://lh3.ggpht.com/_4_R8TkwT74w/Sk...95_Neat_LR.jpg
I initially did a reply to this quote; decided to do a separate thread to attract - hopefully - a wider response than might be limited to a 50D thread.Quote:
The 50D is great for flight Jay. Image quality issues aside, it has the best AF for flight of any Canon body IMO. As Tim points out, you need to get the bird fairly large in the frame for good image quality with the 50D (and with most other cameras for that matter). That's why I'm not a big fan of spending a lot of money on a 300mm lens for birds. In most situations, you are either forced to use a 2x or must crop significantly.
Cheers, Jay
I have to say, I have been pretty satisfied with the 50D. The noise is not bad is you expose to the right. The AF is super fast IMO, and it can produce extremely sharp images, when properly handled. I use a 100-400mm and even at 100% the images look sharp to me. Some of them could maybe be a tad sharper, but I think that is more to do with my glass than the 50D. However as Tim noted, 100% is a huge magnification for the 50D, and that my images are usually very sharp at 100%, I have been very satisfied with the 50D. It certainely blows away my old Rebel XTi.
