Author Topic: Preserving white background through RAW, TIF, JPG? (Read 10293 times)

pet_chia · « **on:** March 22, 2010, 18:21 »

I'm having a problem which is puzzling me. While shooting with my Canon T2i/550d in Raw+Fine mode in the studio, the background is blown out pure white as I intended. In the camera's playback with histogram, the background is blinking to indicate this.

I open the raw CR2 image in Digital Photo Professional v3.8.0 and adjust the camera/lens settings (noise reduction, sharpening, or whatever). In DPP the "View->Highlight Alert" indicates that the background is blown out. So far so good. I do "File->Convert and save ... " and save as "TIFF 16 bit".

Then I open the TIF in Photoshop CS3 to touch up specks of dust, etc. Moving the eyedropper around the background, the info window says:

Quote

R: 255
G: 255
B: 255

8 bit

When finished I do "File->Save for Web and Devices ..." and save with the parameters JPEG/Maximum, Quality: 100, Blur: 0. Here's where the first problem is apparent. Moving the cursor around the background in the "Save for Web" window, the bottom of the screen says:

Quote

R: 255 G: 254 B: 255 Alpha: 255 Hex: FFFEFF Index: --

Why is the green channel now indicating 254?

If I save the file and then open the JPG in CS3, the eyedropper now says that the entire background of the JPG file is:

Quote

R: 254
G: 254
B: 254

8 bit

Weird, huh? Some kind of rounding error converting from 16 bit to 8 bits? Something not working in T2i/550d raw mode, in DPP or in CS3?

I went to a lot of trouble to get the proper studio setup so I wouldn't have to cut out, isolate, brighten the background, etc. in postprocessing, and the software is misbehaving in a way I don't understand. Or maybe I'm misbehaving.

pet_chia · « **Reply #1 on:** March 22, 2010, 18:53 »

I tried using DPP to convert to "Exif-TIFF 8bit" instead of "TIFF 16bit" and the problem no longer happens. The 255/255/255 background stays that way when opened in PS CS3 and converted to JPEG.

I still have a couple of questions -

1) Is DPP messing up the conversion of RAW to TIFF 16bit, or is CS3 messing up the handling of TIFF 16bit? Or is the camera firmware messing up the encoding of RAW?

2) Is there a significant practical reason to prefer 16-bit TIFF to 8-bit TIFF as a middle step between RAW to JPG? I suppose that the 16-bit version preserves more color information, or whatever, but in this case I assume that the "extra" color information is somehow being rounded off, truncated or otherwise converted incorrectly at some stage.

RT · « **Reply #2 on:** March 22, 2010, 19:12 »

Quote from: pet_chia on March 22, 2010, 18:21

......... the background is blown out pure white as I intended. In the camera's playback with histogram, the background is blinking to indicate this.

That's not what it's indicating, the blinking is indicating that particular area of the image is overexposed based on the settings you've taken the image at, that doesn't mean the background is pure white.

madelaide · « **Reply #3 on:** March 22, 2010, 19:56 »

Quote from: RT on March 22, 2010, 19:12

That's not what it's indicating, the blinking is indicating that particular area of the image is overexposed based on the settings you've taken the image at, that doesn't mean the background is pure white.

I always thought the blinking indicated clipped highlights.

Sean Locke Photography · « **Reply #4 on:** March 22, 2010, 20:14 »

Quote from: madelaide on March 22, 2010, 19:56

Quote from: RT on March 22, 2010, 19:12
That's not what it's indicating, the blinking is indicating that particular area of the image is overexposed based on the settings you've taken the image at, that doesn't mean the background is pure white.
I always thought the blinking indicated clipped highlights.

I as well. Higher than 255 at 0 exposure in the raw converter, and that's what I've experienced.

pet_chia · « **Reply #5 on:** March 22, 2010, 20:27 »

Got it ... Playing around more in DPP, I tried tweaking a few more of the parameters away from their default values. The one which is the guilty party is "Auto Lighting Optimizer". Switching it off before saving the 16bit TIFF means that when the TIFF is opened in CS3, the full 255/255/255 white background highlights are preserved when the image is eventually converted to JPEG.

It's a weird problem because it "looks" like it's still 255/255/255 in CS3 according to the image info, but evidently the Auto Lighting Optimizer shaves a tiny amount off the 16-bit RGB values which only manifests itself when the CS3 JPG conversion algorithm decides that 254/254/254 is what they really meant.

Whew!

RT · « **Reply #6 on:** March 23, 2010, 04:51 »

Quote from: sjlocke on March 22, 2010, 20:14

Quote from: madelaide on March 22, 2010, 19:56
Quote from: RT on March 22, 2010, 19:12
That's not what it's indicating, the blinking is indicating that particular area of the image is overexposed based on the settings you've taken the image at, that doesn't mean the background is pure white.
I always thought the blinking indicated clipped highlights.

I as well. Higher than 255 at 0 exposure in the raw converter, and that's what I've experienced.

Cameras as I'm sure you well know don't know that white exists, to a camera a correctly exposed image will have an 18% (or thereabouts) balanced midtone, the highlight alert is to indicate which part of the image the camera thinks is overexposed based on that 18% which it takes from the settings you've applied to take the shot at.

Try it yourself, set up a shot of something white, let the camera auto meter the shot and the result will be that the white object comes out grey, then manually set the camera to overexpose the shot by 1/3 and you'll get the blinking highlights but this is not enough to make the white object 255 it will still be grey albeit a bit lighter, in order to get pure 255 white you'd probably need to overexpose by 1 stop or more.

Sean if you've got your 255 it's because you've correctly overexposed enough to get 255 but you'd still have got the blinking on your camera at 254,253 etc etc if that makes sense.

Something else to consider is that the cameras histogram will not be the same as the image histogram when you go to convert the RAW file, and to be really geeky the cameras highlight alert only blinks if all three channels are 'overexposed'.

To a camera a correctly exposed shot of something white will result in it being 18% grey, an exposure that makes the white thing 16% grey will in the cameras mind be overexposed and it'll blink at you

Personally I don't use the highlight alert, I go by the histogram.

NitorPhoto · « **Reply #7 on:** March 23, 2010, 05:28 »

Is your eyedropper set to 1pix or is it measuring the average of a 3pix or a more wide area?

pet_chia · « **Reply #8 on:** March 24, 2010, 10:01 »

Quote from: NitorPhoto on March 23, 2010, 05:28

Is your eyedropper set to 1pix or is it measuring the average of a 3pix or a more wide area?

It says "point sample" in CS3 when the eyedropper is selected.

microstockphoto.co.uk · « **Reply #9 on:** March 24, 2010, 16:57 »

the weird not pure white problem happened to me as well at times, when saving from tif 8 bit to jpg; sorry I don't have a clue about the reason - it could be a roundoff problem, or a dithering error, or (in my case) a .jpg compression problem but I don't know

regarding your other question:

Quote from: pet_chia on March 22, 2010, 18:53

2) Is there a significant practical reason to prefer 16-bit TIFF to 8-bit TIFF as a middle step between RAW to JPG?

every time you edit a picture a lossy transformation is likely to happen (except for a very few transformations such as 90° rotations / mirror / negative which are lossless and revertible);
every lossy transformation results in errors in the least significant bits; if you do all the editing at 16 bit and only convert to 8 bit at the end, all intermediate roundoff errors are negligible; otherwise, if you do all the editing at 8 bit, you will likely end up with having only 7 or 6 actual bit depth, the rest being just random noise;

it's something similar to mixing music at 24 (or 32) bits before converting to 16 bits for the final cd master

MicrostockGroup Sponsors

Author Topic: Preserving white background through RAW, TIF, JPG? (Read 10293 times)

pet_chia

pet_chia

RT

madelaide

Sean Locke Photography

pet_chia

RT

NitorPhoto

pet_chia

microstockphoto.co.uk

Related Topics

Sponsors

Microstock Poll Results

Sponsors