High ISO Avocet shot

[arash_hazeghi]

This image was made last week in twilight just before sunrise, it was actually much darker than the photo shows, I like very early morning light, just before sun is up. It is dim but very uniform and perfect for those blacks and whites with no shadow. Do need to be careful with WB to prevent a blue shift, will cause more color distortion/noise if you attempt to correct in post!



1D4 500 +1.4XTC, f/5.6 1/1600sec ISO 6400. sw DPP.

Comments welcomed.

[Lance Peters]

Hi Arash - Nice - I am continually amazed how good 6400 can be on the current gen camera's - to get really good usable results at 6400 ISO is just fantastic.
Like this a lot - nice work.

[Arthur Morris]

Are you saying that changing the WB during conversion (rather than in-camera) will cause problems? Nice image BTW.

[Dave Mills]

I'm also impressed with the quality at 6400. Good comp with nice detail throughout. I like the fact your showing the environment and yet it's not distracting. Well handled!

[Cory Bucher]

WOW, 6400 looks really good. Like the pose and the surrounding environment. If any what noise reduction software did you use?

[Michael Pancier]

I too am amazed as to how clean this is. Curious if you used NR as well.

[Bill Dix]

Arash, lovely shot in low light -- one more in a superb string of Avocets. I agree with Dave on the bg - just the right amount of oof to show context w/out distraction.

[arash_hazeghi]

Are you saying that changing the WB during conversion (rather than in-camera) will cause problems? Nice image BTW.

Thanks :) in high ISO it does, most RAW converters will pull up the other channels which will enhance the appearance of noise and nasty color blotches. becomes tough to remove.

[arash_hazeghi]

WOW, 6400 looks really good. Like the pose and the surrounding environment. If any what noise reduction software did you use?

I use Canon DPP software for RAW conversion so I get cleaner output to begin with. then I applied one round of noise reduction to the BG only with NeatImage plugin for Photoshop. I did not apply external NR to the Avocet, the white feathers will not have any visible noise at small size and for dark feather I set the shadow fade amount in smart sharpen to 15 so darks are not sharpened as much.


Here is a 100% crop from DPP, you can see most of the noise is in BG, furthermore the grain size for DPP output is very small compared to actual details, in other words noise and detail are on different spatial frequencies, NeatImage lets you apply different amount of NR for different frequencies you can then exploit this to remove the fine-pitched noise without affecting detail much.

[Ian Cassell]

Hmmm .... I'm reading this with interest.

It was my understanding that RAW converters were just that -- and didn't process the image. That's not true?

So, Arash, is DPP a better RAW converter than the current versions of ACR? I essentially never use DPP, but it wouldn't be such a big deal to switch.

[arash_hazeghi]

Hmmm .... I'm reading this with interest.

It was my understanding that RAW converters were just that -- and didn't process the image. That's not true?

So, Arash, is DPP a better RAW converter than the current versions of ACR? I essentially never use DPP, but it wouldn't be such a big deal to switch.

Ian RAW converter does make a big difference in the final output when it comes to high ISO!!! I prefer DPP over ACR. You have to try them all and see which one you like best. here is a poll from before shows what other people are using:

http://www.birdphotographers.net/for...are-do-you-use

[Ian Cassell]

That IS interesting. Does this hold with the newest iterations of ACR? I usually import with ACR into LR3 to maintain my catalog of images, but do most of my PP in CS5. Then I use Noiseware Pro plugin. I see that DPP has a "Transfer to Photoshop" selection in tools, but I don't see a way to dump RAW output from DPP directly into LR3. If I switch, this is going to muddle my workflow more than it already has been muddled. Do you use DPP for everything or only for your problem high-ISO files?

I've also always done my WB adjustment in PP -- don't recall ever having set it on-camera. I guess this is what I get when I "assume".

[arash_hazeghi]

That IS interesting. Does this hold with the newest iterations of ACR? I usually import with ACR into LR3 to maintain my catalog of images, but do most of my PP in CS5. Then I use Noiseware Pro plugin. I see that DPP has a "Transfer to Photoshop" selection in tools, but I don't see a way to dump RAW output from DPP directly into LR3. If I switch, this is going to muddle my workflow more than it already has been muddled. Do you use DPP for everything or only for your problem high-ISO files?

It is true even for the latest version IMO, it is not surprising because Canon DPP uses propitiatory algorithm for demosaic and NR which take into account the detailed characteristic of the image sensor, Adobe do not have this information so they use generic algorithms. You cannot transfer a "RAW" file between applications, a RAW file is just RAW data and every application massages it in a way that is different, the adjustments you make by DPP are not readable by ACR or vice versa so you cannot transfer a RAW file. If you use DPP then you can transfer the TIFF output to photoshop (that's what transfer to photoshop does) and continue editing there. I use DPP for all conversions. Thanks

[Michael Gerald-Yamasaki]

Are you saying that changing the WB during conversion (rather than in-camera) will cause problems? Nice image BTW.

Thanks :) in high ISO it does, most RAW converters will pull up the other channels which will enhance the appearance of noise and nasty color blotches. becomes tough to remove.

Arash,

Greetings. Hmm... WB isn't expressed in the RAW Bayer data, but in the metadata (for use by the converter). Shooting in one WB has no impact on the conversion to another (AFAIK ;)). The noise and nasty color blotches are more likely from the raw conversion "curve" used in mapping Bayer to RGB. In Nikon Capture NX2 this curve is available for adjustment using Picture Controls (and the Picture Controls Editor) or picture controls in camera... Canon/DPP I don't know. The initial raw conversion curve has significant influence on the expression of noise, but this too is adjustable out of camera.

Great avocet shot... the color and detail are superb.

Cheers,

-Michael-

[arash_hazeghi]

Arash,

Greetings. Hmm... WB isn't expressed in the RAW Bayer data, but in the metadata (for use by the converter). Shooting in one WB has no impact on the conversion to another (AFAIK ;)). The noise and nasty color blotches are more likely from the raw conversion "curve" used in mapping Bayer to RGB. In Nikon Capture NX2 this curve is available for adjustment using Picture Controls (and the Picture Controls Editor) or picture controls in camera... Canon/DPP I don't know. The initial raw conversion curve has significant influence on the expression of noise, but this too is adjustable out of camera.

Great avocet shot... the color and detail are superb.

Cheers,

-Michael-

Michael, I think you missed the point, of course WB does not affect raw capture but it affects the curve that is applied. WB adjustment is done by changing the curves that are applied to form each of the RGB channels during demosaic and as such it does affect the appearance of color noise.

Also Canon and Nikon are not equal in terms of color noise, Nikon subtract color noise on-chip so you cannot compare the two. If you own a Canon camera you can set it to ISO 3200, or 6400 then pick florescent WB and shoot a dim scene, then open the file in DPP and click on a neutral point for WB and see how nasty it will look like :)

[Arthur Morris]

Not sure who my money is on :) I guess the best thing to do is to try it as Arash suggests (but possibly not with such a dramatic shift....) Interesting discussion that points to the word on the street being wrong.

[arash_hazeghi]

Arite,
Have you made any images at ISO 6400 or more with the MK4? can you post some?

[Arthur Morris]

Have not yet. Been blessed with pretty good light. Will be on the lookout for the right situation. Made some good ones with the MIII at 3200 and 6400.

[arash_hazeghi]

For those who are interested in high ISO technique this image was made with the "old" 8 mpixel canon 20D at ISO 3200 (highest it could go) , it has also been cropped. with the same noise reduction technique discussed above, this one had noise reduction applied to the bird as well.

[arash_hazeghi]

Have not yet. Been blessed with pretty good light. Will be on the lookout for the right situation. Made some good ones with the MIII at 3200 and 6400.

Don't hesitate to try, they have made improvements in MK4 and it is better than MK3, works well up to 6400 if lightning is uniform. MK4 also handles accidental underexposure well:D I am going to print this at 18X12 and see how it will look, my guess is no one will be able to tell it was 6400.

BTW the avocet image is not even FF, it is like a 60% crop, details hold well. it's all about processing :)

[Arthur Morris]

Do you use DPP only for your high ISO images?

[arash_hazeghi]

Do you use DPP only for your high ISO images?

No I use it all the time, like the quality and speed. I know I am against the tide but I actually consider it one of the advantages of Canon system, plus not having to constantly buy and upgrade LR/Adobe Photoshop saves enough money to buy new lenses every once in a while :D

[Ian Cassell]

So, Arash, if I'm hearing you correctly ... you set WB in-camera specifically for each situation?

[arash_hazeghi]

So, Arash, if I'm hearing you correctly ... you set WB in-camera specifically for each situation?

No Ian, I use AWB 99% of the time, when I use high ISO (3200 or higher) in conditions that AWB will cause a strong cast in the image (e.x. early morning light or very late afternoon) I use custom WB for better results.

[Harshad Barve]

I am following this thread , excellent discussions , BTW lovely image Arashbhai
TFS

[Harshad Barve]

For me raising ISO beyond 400 is like writing off my property but would like to know , how BIG I can print such high ISO images ,

[arash_hazeghi]

For me raising ISO beyond 400 is like writing off my property but would like to know , how BIG I can print such high ISO images ,

what camera are you using?

This one you can easily print up to 18X12 if processed correctly, fine grain will not show in print.

[Alfred Forns]

Sweet threat Hope everyone reads over it twice, lots of good information !!!

[Robert O'Toole]

.........Do need to be careful with WB to prevent a blue shift, will cause more color distortion/noise if you attempt to correct in post!

Nice image and the good work on the NR at 6400!

Is it just me or do I see a bluish cast in the whites, I just had my monitor calibrated last weekend so I on the look out for casts. Or is the blue cast something you tried to achieve?

Anyway I saved and opened in PS CS5, the blue channel is showing a cast. I neutralized the cast for you. I think the non blue cast version looks better but not sure what you were trying for when you made the image.

I have a great way to set custom WB in the field that I teach in my workshops and I can share if you are interested.

Robert

[Ian Cassell]

I have a great way to set custom WB in the field that I teach in my workshops and I can share if you are interested.

Robert

Please do! Thanks.

[Robert O'Toole]

Michael, I think you missed the point, of course WB does not affect raw capture but it affects the curve that is applied. WB adjustment is done by changing the curves that are applied to form each of the RGB channels during demosaic and as such it does affect the appearance of color noise....

Good topic, this topic has been covered on BPN in the past actually.

Arash is right.

But some people might need another simpler explanation. WB does affect noise in a converted image in the same way even a color space setting would. Since you are looking at an embedded JPEG histogram to set your exposure in manual or your exposure comp in the auto modes WB will have a huge effect on the final image noise level.

The problem that comes into play is that most, nyself included, trust that the histogram is correct when in reality it is not. The blue and red channels are boosted for display to balance the green channel since with a Bayer layout there are 2X as many green photosites as R and B. You can test this by shooting a gray card with custom WB then looking at the curves in the image. They are not all equal nor does the histogram of the image match the histogram on the back of the camera. To correct for this you have to use a custom WB that will give you greenish images.
Jim Fenton has used this technique in the past with success. It is called UniWB.

I am one of those crazy people that like to set custome WB settings in the field when quality and the color of the light is very important. In fact I teach this in the field when I am working leading workshops. I can share it with everyone if you are interested.


Hope this helps.

Robert

[Ken Lassman]

Arash my friend....you keep impressing me with your technical knowledge. Very impressed with the high ISO pic. Are you no longer using Nikon ?

[Robert O'Toole]

Please do! Thanks.

Okay Ian, nice to know someone out there is interested!

I dont want to hijack the thread so I will write something up later and post somewhere and include the link in a reply here so you can find it.

Robert

[Ian Cassell]

Thanks! I'm sure I'm not the only one who would be interested.

[Jackie Schuknecht]

Very interesting thread, understand all except the Bayer part. Hope you will post something Robert. Great info and Arash you blow me away with these high ISO images. You do have a wealth of technical info, much more so than the average 'tog. TFS.

[Dan Brown]

Okay Ian, nice to know someone out there is interested!

I dont want to hijack the thread so I will write something up later and post somewhere and include the link in a reply here so you can find it.

Robert

Thanks Robert! I am very interested also!

[denise ippolito]

Arash, Thank you for posting this. You are so helpful to all of us with these informative posts. Please keep them coming. Looking forward to your next one.

Robert, I am interested also. Thank you for sharing.

[arash_hazeghi]

Arash my friend....you keep impressing me with your technical knowledge. Very impressed with the high ISO pic. Are you no longer using Nikon ?

Hi Ken, I no longer have Nikon gear and I do miss it (D700). I still have lots of nice files from before I sometimes play with in NX2.

[Desmond Chan]

Is it just me or do I see a bluish cast in the whites,

I see the cast, too.

Great thread nonetheless !

[arash_hazeghi]

I see the cast, too.

Great thread nonetheless !

That is not a WB cast, since WB was adjusted manually using a gray card the colors reflect the accurate colors of the scene at the time photograph was made. The photo was made in twilight before sunrise and I prefer the colors as close as they were in real life.

[Robert Park]

Arash
First off this is my first post here and I was compelled to join largely by revelation on this thread. I was extremely skeptical about WB effecting raw exposure.
I did a test based on your findings. I set iso to 1600 and made 3 exposures in AV mode with the camera on a tripod so the scene was identical.
1 exposure at AWB
1 exposure at Tungsten wb
1 exposure at shade wb
I opened them up in ACR 6 and synched all to the 4700 k that the awb calculated.
The histograms have some definite difference and some clipping is going on in the highlights that differs on each.
The images have slightly different looks in the shadows too.

Holy smoke I never would have thought and at very high ISO there would be even more clipping going on with the reduction in DR.
I need to think more about this and how it will affect me down the road.

[Robert O'Toole]

That is not a WB cast, since WB was adjusted manually using a gray card the colors reflect the accurate colors of the scene at the time photograph was made. The photo was made in twilight before sunrise and I prefer the colors as close as they were in real life.

Arash are you saying that you did apply a custom WB to the first image with the blue cast using a gray card as a reference? If this is so this is incorrect .

This is a a very basic digital photography technique. If you use a neutral gray card to manually set a custom white balance properly you will eliminate any color cast.

If you did in fact use a neutral gray card and applied a custom WB using the gray ref image then it must have been done incorrectly or you fine tuned the WB setting to introduce the blue cast. A properly WB image using a gray card will not have a cast.

Second using a gray card to make a custom WB to get accurate colors in early morning light is just plain wrong. Using a gray card to make a custom white balance will removed the color cast from the image.

Either way if you like the blue cast in your image that is your preference but when you tell people to be careful about avoiding a blue shift in low light then you post an image with a blue cast its counter productive unless you mention that you left the image with a blue cast intentionally. Then you say you used a gray card as a reference to make the image blue and that is correct?

Please explain.

Robert

[Robert O'Toole]

Robert, I am interested also. Thank you for sharing.

No free time this weekend to start the thread I mentioned but I will get to it monday or tuesday.

Robert

[Robert Park]

One should remember the psychological effect of color and image processing. There are 2 images posted above, one original and one that has been warmed.
The mind looks at the original and interprets a low light dawn or dusk setting. The clues are taken from the apparent contrast both local and global.
The second image looks out of place with just a color balance change. It has a daylight balance but not the proper daylight contrast structure.
One extreme example is a night image that has been fully color corrected to appear like daylight. They just seem wrong to the mind and are interpreted
as different that what daylight should be.
I personally think that if you are going to color balance the image it should be less warm that the suggested fix and have more contrast structure.
I for one use the 5500k as my standard for my work and prefer to use the natural color shifts in shadow or twilight as a creative tool.

[arash_hazeghi]

Arash are you saying that you did apply a custom WB to the first image with the blue cast using a gray card as a reference? If this is so this is incorrect .

This is a a very basic digital photography technique. If you use a neutral gray card to manually set a custom white balance properly you will eliminate any color cast.

If you did in fact use a neutral gray card and applied a custom WB using the gray ref image then it must have been done incorrectly or you fine tuned the WB setting to introduce the blue cast. A properly WB image using a gray card will not have a cast.

Second using a gray card to make a custom WB to get accurate colors in early morning light is just plain wrong. Using a gray card to make a custom white balance will removed the color cast from the image.

Either way if you like the blue cast in your image that is your preference but when you tell people to be careful about avoiding a blue shift in low light then you post an image with a blue cast its counter productive unless you mention that you left the image with a blue cast intentionally. Then you say you used a gray card as a reference to make the image blue and that is correct?

Please explain.

Robert

Robert I know what I am doing when I said I used a gray card. I don't like your repost as the color balance is totally off. I am also aware of the basics of digital photography and don't really have the time or the interest to go back and forth. If you don't agree it's fine but I am going to leave it here-to each his opinion.

Thanks everybody for comments and feedback.

[Robert O'Toole]

Robert I know what I am doing when I said I used a gray card. I don't like your repost as the color balance is totally off. I am also aware of the basics of digital photography and don't really have the time or the interest to go back and forth. If you don't agree it's fine but I am going to leave it here-to each his opinion.

Thanks everybody for comments and feedback.

The only reason I asked for further details was for the benefit of others on BPN. The the main goals or BPN are to share and learn right?

It would have been more helpful for the readers of this thread if you could have just typed a line or two to explain rather than say bad things about my repose as I was only trying to help.

Robert

[Arthur Morris]

Robert I know what I am doing when I said I used a gray card. I don't like your repost as the color balance is totally off. I am also aware of the basics of digital photography and don't really have the time or the interest to go back and forth. If you don't agree it's fine but I am going to leave it here-to each his opinion. Thanks everybody for comments and feedback.

Hi Arash, While I have not had time to study this post in detail, I will say that your originally posted image does have a blue cast to the whites (as viewed on three different calibrated monitors). In addition, placing the cursor on any of the whites show the distinct blue cast. On the brightest whites the RGB values are 240, 240, 252. As far as I know, there is no arguing with that.

You might have tried to state your position here "I .... don't have the time or the interest to go back and forth" with a bit more civility. We would appreciate your doing so in the future.

[Arthur Morris]

One should remember the psychological effect of color and image processing. There are 2 images posted above, one original and one that has been warmed. The ind looks at the original and interprets a low light dawn or dusk setting. The clues are taken from the apparent contrast both local and global. The second image looks out of place with just a color balance change. It has a daylight balance but not the proper daylight contrast structure.
One extreme example is a night image that has been fully color corrected to appear like daylight. They just seem wrong to the mind and are interpreted as different that what daylight should be. I personally think that if you are going to color balance the image it should be less warm that the suggested fix and have more contrast structure. I for one use the 5500k as my standard for my work and prefer to use the natural color shifts in shadow or twilight as a creative tool.

Robert. Your thoughts above are all correct. One's post-processing choice of color balances are of course 100% personal. Thanks for sharing them. I do believe that you are missing Robert O'Toole's point here. He simply stated that the whites in the image in Pane 1 have a blue cast. He is correct. Arash maintains that since he used a gray card that the image does not have a color cast; the fact is that the bright whites in that image are R: 240, G: 240, and B: 252. That is a blue cast.

[arash_hazeghi]

Hi Arash, While I have not had time to study this post in detail, I will say that your originally posted image does have a blue cast to the whites (as viewed on three different calibrated monitors). In addition, placing the cursor on any of the whites show the distinct blue cast. On the brightest whites the RGB values are 240, 240, 252. As far as I know, there is no arguing with that.

You might have tried to state your position here "I .... don't have the time or the interest to go back and forth" with a bit more civility. We would appreciate your doing so in the future.

Artie, with all due respect, I am aware that Robert Works for you and you are close friends but please first study the thread and some of the history. I feel like you just came to defend him, which I understand but Robert OT has made inappropriate comments on my photos in the past, this is a public forum so everyone is free to comment on the photos, however whether I see a comment on my photos worthy of response and choose to reply to it is at my discretion. I think you do the same.


Re: the cast, if you read the thread I mentioned that this image was made in early morning, the main source of light in the image was not sunlight but dim ambient light reflected from the water in the pond, this light obviously has a blue spectrum, by simple laws of physics a white object will reflect the blue light and therefore look blueish. Setting the neutral point on the Avocet feathers will throw off the color balance and the image will not look natural, as pointed out by Robert Park. Subtracting blue has caused a strong yellow cast on the greens that was not there. Definition of color cast as far as I know is an unwanted or unnatural color tint which is unfaithful to the subject and the scene, in this case the first image is as close as possible to the original scene which is the way I prefer...

Best and thanks for comments and participating.

[Dan Avelon]

I have been following this thread too, thanks for the info about processing high ISO photos, before I was always afraid of using anything above 800 but looks like with good post processing you can get great results.

I also live in the San Jose area, this must be from Palo Alto Baylands or the Shoreline park. I sometimes go there early in the mornings too, I can say that the original post is exactly how it looks like there, the marshes in the BG are rendered so nicely. the 2nd image looks unnatural to me I would stick to the original. I am going to install DPP and neat image and start playing with them. It is great that you share your knowldege and tricks with everyone for free!



Cheers,
Dan