Canon has announced EOS 7D

[Flavio Rose]

Responding to John Crookes, I just wanted to note that your first still life image looks to me very smeared on both sides -- much more so than the imaging-resource ISO 200 samples for either the 7D or D300. Especially the second handkerchief from right looks very smeared.

Technical details: I am looking at your post with Firefox 3.5.2. The image in your post looks the same when I download it and open it in Gimp and with Windows XP's Picture and Fax Viewer. Gimp converts your downloaded image from ProPhoto RGB when I open it. In contrast, the images at imaging-resource appear to be in sRGB since Gimp gives me no conversion message. Could it be that your screen capture process somehow damaged the image?

[arash_hazeghi]

Imaging resource samples are shot in JPEG mode, the camera's DSP reduces the noise by softening the photo, RG photos are in RAW mode. I am really surprised that so many people here do not realize the difference between RAW and JPEG formats. Any ways, whatever works for you :D

RG's comments on AF and his sample images (especially Thunder and Bird) leave me dismayed. For a fair comparison I've looked at the images at a 75% linear reduction so the pixel density is about same as on my 10 megapixel XTi. The ISO 400 still doesn't look any better to my eyes than what I get from the XTi.

I try to rationalize the AF comments by saying that RG didn't like the 1D Mark III's AF, and Canon would never give a non-1D camera AF better than the current 1D Mark X, so it was hard to expect that he could like the 7D's AF. Still...

I am mystified why RG's ISO 400 images look perceptibly worse than imaging-resource's, which were the first ones I looked at. I realize that imaging-resource exposes to the right in its studio setting, while RG may not have done so in outdoors shooting. One can only hope cynically that maybe Canon sent him a prototype with bum AF and a bum sensor (a) as a reward for his past comments and (b) in the hope he'd write bad things about the 7D that would prove false.

[john crookes]

Arash

The samples that galbrath display are raw file formats and they are even noiser.

Imaging resources displays jpegs to save on space and also to compare same results as you can not show a raw file but only a file that has been converted somehow

[Doug Brown]

The reds on the 7D image are oversaturated compared to the D300. Pull them back a bit in PS and this is what you get. I now see more detail in the 7D image.

[Flavio Rose]

Responding to Arash, RG photos are in raw mode but he says that they were noise reduced in DPP 3.7. What you seem to be suggesting is that imaging-resource's 7D applied a much harder noise reduction in camera than the noise reduction Galbraith says that he applied in DPP 3.7. That is a possible explanation for the difference in the images. I would have imagined that the DPP 3.7 software noise reduction (NR) algorithm would be similar to the Canon 7D firmware NR algorithms, but you never know, particularly since you don't know where Galbraith set the NR sliders in DPP relative to what the camera does.

[Doug Brown]

Actually John you remember a Mallard photo. I just processed this one for kicks tonight. I believe they were both taken on the same day when I was experimenting with high ISO settings on the 50D.

[arash_hazeghi]

Responding to Arash, RG photos are in raw mode but he says that they were noise reduced in DPP 3.7. What you seem to be suggesting is that imaging-resource's 7D applied a much harder noise reduction in camera than the noise reduction Galbraith says that he applied in DPP 3.7. That is a possible explanation for the difference in the images. I would have imagined that the DPP 3.7 software noise reduction (NR) algorithm would be similar to the Canon 7D firmware NR algorithms, but you never know, particularly since you don't know where Galbraith set the NR sliders in DPP relative to what the camera does.

Falvio,
If you are familiar with Canon EOS system you know that Canon DPP is the best RAW converter for CR2 files from any EOS body in terms of IQ alone. The output from DPP never matches that of in-camera JPEG for any EOS camera, the algorithms are different and much more complicated in DPP. Show me one EOS camera which produces a JPEG file that is the same as DPP. You can test this if you have a Canon body right now.

Imaging resource samples are heavily noised reduced and do not show that much detail both for Canon and Nikon, RG knows what he is doing so he uses DPP and applies just a little bit of NR to get rid of color noise and squeeze all the detail. As such the camera is noisy, there is nothing you can do. Laws of physics and electronics that govern the operation of CMOS image sensor demand so. I was actually engaged in another thread here that was about sensor noise and issues and I am not going start another one here, All Canon/Nikon/Sony do is that they use sophisticated DSP to deliver a clean processed JPEG albeit with low detail out of the camera, this does not apply to me who shoots in RAW.

[arash_hazeghi]

There is no question 50d can deliver good images at high ISO in good light once resized and NRed and when bright feathers can hide noise, here is an image taken with 50D at ISO 1600, has lots of detail





But when light drops then noise kicks in,



50D ISO 3200



Ooops, no more feathers :(

Also comparing JPEG is a waste of time, Here I downloaded 7D ISO 100 JPEG sample from IR website opened in PhotoShop, then I downloaded the same scene taken with Nikon D700 at ISO 200 but in RAW note that same Sigma lens was also used for both cameras, converted the NEF file with my default settings in Light Room and then upsized it to 18 mpixel size to stretch it to the max and put it side by side to EOS 7D. Which one is sharper now? Look at the fabric texture, it is gone in Canon shot.



If you want to compare two files you need to download RAW files convert them with ACR and then put them side by side.

[Flavio Rose]

Thank you Arash, particularly for the last image which I was able to compare with imaging-resource's jpeg and so see for myself the D700's default noise reduction in action at ISO 400.

Ironically I shoot mostly raw on my XTi, so my mental image of what ISO 400 should look like comes from seeing my raw files when I first open them in DPP, before NR. It's easy enough, I guess, to switch the camera to raw+jpeg for a little while, something I don't recall ever doing (!), and compare the camera's jpeg with DPP's to see how much smearing the camera's algorithms perform by default.

In DPP I have often wondered if zero is really zero on the noise reduction sliders, or if there's some inevitable level of NR you get. I've also wondered what slider positions match what the camera does by default, but since I'm shooting mostly raw I didn't much care. You seem to be saying that there are no such slider positions because the algorithms are really different.

[arash_hazeghi]

Thank you Arash, particularly for the last image which I was able to compare with imaging-resource's jpeg and so see for myself the D700's default noise reduction in action at ISO 400.

Ironically I shoot mostly raw on my XTi, so my mental image of what ISO 400 should look like comes from seeing my raw files when I first open them in DPP, before NR. It's easy enough, I guess, to switch the camera to raw+jpeg for a little while, something I don't recall ever doing (!), and compare the camera's jpeg with DPP's to see how much smearing the camera's algorithms perform by default.

In DPP I have often wondered if zero is really zero on the noise reduction sliders, or if there's some inevitable level of NR you get. I've also wondered what slider positions match what the camera does by default, but since I'm shooting mostly raw I didn't much care. You seem to be saying that there are no such slider positions because the algorithms are really different.

Flavio a few points

1) Older than Canon cameras did not have aggressive NR as the new ones today, ironically the number behind the digic processor also indicates strength of NR :D , my old 20D had very little NR even in JPEG mode, 40D had more, 50D has so much that I never bother with JPEG.

2) If you want to replicate in-camera JPEG you have to use EOS Viewer utility (previously supplied with Canon DSLRs) or older versions of Zoom Browser that included RIT (Raw Image Task which they have now discontinued) to convert the CR2 files.

3) DPP has two slider for NR, one is chroma which essentially gets rid of color blotches without affecting detail, I always use it and it does its job nicely. The other slider is luminance NR, it filters high spatial frequencies (noise and detail alike) to provide a smoother image, it can destroy feather detail I never use it.

4) Experienced shooters like RG have a very optimized work flow, most likely he uses chroma NR in DPP and no luminance NR as his photos are very sharp, I have seen his samples before and I trust him. I can't get any better than his photos. At the end of the day if you can get good photos, even with NR it is ok and who cares how you got there. But sometimes noise is so much that you can't do anything. Plus I personally prefer to spend the little extra time I have for photography spent outdoors taking pictures, not running NR on the computer, as for I already deal with enough all noise sorts in the lab everyday!!!!

BTW, regarding your question about sensor circuit in the other thread, did you see this article that I posted? This is very useful information everyone interested in camera and sensor should read so I am posting it again here http://micro.magnet.fsu.edu/primer/digitalimaging/cmosimagesensors.html

[Morkel Erasmus]

some interesting points made and the pictures really don't show the high difference in noise levels IMO.
I for one am anxious to read the in-depth reviews but can foresee that I will definitely replace my 1000D with the 7D once it has arrived and the dust has settled...
technologically this looks like an amazing camera for the price point.

if my images @ ISO 3200 can look anything close to Doug's duck I will be a happy man.

[Christopher C.M. Cooke]

Ouch!! What if they called it a 60D? Mate, would that make you happy? Does the label really matter? The 7D does have - on paper - a lot of cool new features. :D

I am with you Jay, pre ordered mine with Michaels in Melbourne only number 2 on the list.

This late in life, I ain't prepared to wait until I can't remember my name before I try it.

Plus I already have two spare 5DMKII batteries, which I have yet to need to use.

[David Fletcher]

If it cannot AF at f/8 then they should have called it a 60D! Canon can stick this where the sun don't shine AFAIAC. Waste of time IMHO. :D

I do like that view Akos!!!!!!! Made me smile. Seems to me the same problems exist with Canon, is that they are "pumping steroids" to use a phrase at pixels and not dealing with their other failings. Some will love it, some won't but I won't be peeking over my shoulder wondering if I was impulsive by switching in 2007 after giving up on the 1DIII.

Thanks Arash for that comparison. Bit scary how the detail has gone in the Canon file at ISO 100.

[Jamie Strickland]

I think the 7D looks fairly noisy, but I suspect that will be fixed with firmware

anyone remember the first pics released from the 5D MII on Canons site? they did not look that great at all but once it was officially released I was blown away by them

lets hope I am right :) otherwise my order will be returned to amazon once I get it

[Pao Dolina]

What are you folks most interested about the 7d?

[phillane]

Jamie

Bottom line I hope the camera exceeds yours and everyone's else's expectations however I will sit and wait with moderate skepticism. My Canon 40D had numerous focusing problems which required several trips to Canon's facility. The issue was never resolved, as a matter of fact Canon took the position that the camera was functioning properly when it was not. At the time a simple search on google using the phrase (Canon 40D focusing problems) yielded a plethora of similar complaints. It's the hide your head in the sand attitutde from Canon that really puts a fire under my #$%@.

[Pao Dolina]

Any news whether Artie will upgrade from a 50D to a 7D?

[Michael Wolf]

Iteresting tidbit.. for me anyway.

According to PEN Vol.30 N032

With this new strategic model, Canon intends to pursue Nikon, who got a 10.4% share with the D90 alone in the DSLR domestic market for the first eight months of this year according to marker data compiled by BCN research company. The 18MP 7D will be sold for around 190,00 yen (US $2,000) body only vs the D300's 230,00 yen ($2,440) and will be produced 60,000 units per month as the same with the Nikon D300.

I also love the fact that you will be able to take full 18MP RAW images while in the middle of shooting HD video!

[Greg Schneider]

Artie, when are you going to have a 7D preview for us? :)

[Jay Gould]

The bird image at the beginning of this thread is pretty awesome: http://photography-on-the.net/forum/...=752819&page=7

As are these reviews:

http://www.naturfokus.info/EN/

http://jeffascough.typepad.com/jeff_...perpignan.html

[Flavio Rose]

The author of the bird image Jay points to, liquidstone, has also stated on dpreview that his very limited AF testing showed less ability to track birds against busy background than the 1D Mark II. Liquidstone says however that he wants to reserve judgment until he gets more time with the 7D.

A further data point is that the 7D has two levels of user-settable AI Servo tracking sensitivity, versus four on the 1D machines and the D700 (and zero on the 40D and 50D). Authoritative BPNers recommend the lowest of the four levels of tracking sensitivity in the 1D's. We do not know where the effect of the 7D's two levels lies relative to the 1D machines' four or indeed whether such a comparison is even meaningful. I would guess that tracking sensitivity is part of a complicated algorithm to eliminate "outliers" from the input to the AF's prediction algorithm. I would guess that the input includes a series of (time, focus point, subject distance estimate, quality of subject distance estimate) records.

Hopefully real reviews, rather than these short previews, will be coming out later in our autumn/Australia's spring:).

[Doug Brown]

I'm expecting mine in the first batch of deliveries and FWIW I plan to do some exhaustive field testing. I will post my findings here on BPN and also on my website.

[arash_hazeghi]

LR 2.5 has beta support for 7D CR2 files now, you can convert the samples on IR website through your standard workflow and inspect image quality.

Here is a comparisons between 7D and 50D ISO400 studio scene through my standard LR workflow, noise is visually identical to 50D with marginally more subtle detail (ex texture area), keep in mind these shots were on tripod with prime macro at f/8. This also shows how poor the JPEGs straight out of the camera are, compare the output from RAW below to the JPEG in post #58. all the texture detail that was lost in JPEG is now recovered in RAW.