A little confused about how ff handles linear images

What are your settings in Tools > Options > Gamma?

That may not have made sense so here is a grab of my settings. You can notice at the top that ff is displaying the message that the image was loaded without correction - with the exr image shown on the right.

Here is the same with it enabled.

I am so confused right now. Here are my cs5 RGB workspace settings.

Here are the results as they appear in cs5. You can see in the upper left corner of each image which gamma they were saved as. Starting form the upper left and going clockwise, 1, .45, 2.2. This is with soft proof off.

Fwiw, all of these images were saved as 32bit exr with alpha and no compression.


Ok, so that seems to be all working as expected. Ps should open and display them according to the gamma that ff saved them with.

(Sorry for the lack of replies -- we'll reply when we have a free programmer to look into this. Everybody is too busy working on Beta Stage 4.)

FF behaviour in your tests is absolutely right.
For example, you save 0.5-grayscale pixel in FF without gamma correction. What will you have in file? Of course 0.5. Then you load previously saved 0.5-pixel without gamma correction. So after loading you get the same pixel value as before saving. No surprises.

Photoshop performs pixel loading in the different way. It supposes that saved in FF picture was created with linear-gamma device. So PS makes linear-monitor
gamma-correction. It's produced according to the formula:
loaded_pixel = pixel_in_file ^ (1 / monitor_gamma)
Actually FF does the same during loading if gamma correction is turned on.


BTW the formula explains why linear-gamma images look washed after loading if they were created with non-linear gamma devices (such as your monitor =).

Quote
ronviers wrote: Filter forge is displaying the images exactly the same if gamma correction is turned on or off.

Only in case of displaying, but saving/loading process in 2 gamma modes is different.
In case we use no gamma correction FF acts as I previously wrote.

In case we use gamma correction, FF performs monitor-linear correction before saving and linear-monitor one before loading.

For better understanding and consistency between PS and FF try the following test: save an image in FF without gamma correction, then turn correction on and load previously saved image. You will get "washed out Photoshop" result.

I hope this explanation is enough. =)

Quote
ronviers wrote: If ff’s gamma correction setting is set to disabled and I open an image that had been saved with ff’s gamma correction setting set to disabled, I see an image that appears to be corrected to monitor but this correction is preview only – the image is linear. Is that correct?

No, it's not. =(
While PS corrects pixel values only for on its previews, FF changes them everywhere as it doesn't have split pipelines for processing and displaying of the image data.
For example, you have saved 0.5-value pixel (1.0 - white, 0.0 - black) in file with linear gamma. After loading in PS you get 0.5 value (check it out with color picker). But pixel value for PS preview is greater, about 0.7. So PS preserves original pixel values for image processing and have a copy of them for gamma-corrected preview.
If you load the image in FF (gamma correction is turned on), you will see the same preview as in PS, but pixel value will be 0.7 (check it out with color inspector).
So FF with enabled gamma changes pixel values during image loading. Of course, FF doesn't change values of the original image, it just copies image into internal cache and produce gamma correction for it.

Why is it so? Besides of image processing FF uses pixel data for numerical calculations. For example, linear-gamma gradients will be really confusing for users who use it for grayscale inputs mapping.