how to fix color issue in slides

[Rasaa84]

Hi @jcupitt
I scanned a slide with the VENTANA DP 200 scanner. When I read the slide with pyvips, the image appears somewhat darker. I’ve already tried applying icc_transform, and I’m using pyvips 3.0.0. Below is the image I obtained after running dzsave.

and here is the code I use:

image = pyvips.Image.new_from_file("slide.tif")
image_srgb = image.icc_transform("srgb")
image_srgb.dzsave("dzi_slides", tile_size=2048)

[jcupitt]

Hi @Rasaa84, could you share the original VENTANA slide, plus a screenshot of the appearance you expect?

[Rasaa84]

This is what I expect to see

and this what I see after icc_transform

[Rasaa84]

would you like me to share the original slide here?

[jcupitt]

Yes, the original slide please, so I can see what openslide is doing. Something like a google drive link would be great.

[Rasaa84]

@jcupitt
here is the link to the original slide https://drive.google.com/file/d/1zqQ6RRLREOS3mnvh47x_e4mK_1BsVBUH/view?usp=drive_link

[jcupitt]

Thanks!

Your original image has no colour profile, so image.icc_transform("srgb") won't do anything, unfortunately. The ventana image viewer (is that what your screenshot is from?) must be doing some other, internal colour adjustment.

You need to compare the expected result to the result from openslide and reverse-engineer this adjustment. Probably (probably!) all you need to do is remove gamma correction from the two images, compare them to estimate the adjusted primaries, generate a 3x3 matrix, and reapply gamma. This should let you compute an ICC profile which will give you a similar result.

The ventana viewer might be doing something more complicated, but I'd expect just adjusting the primaries would be good enough.

[Rasaa84]

when I run this

image = pyvips.Image.new_from_file("path/to/slide", access="sequential")
icc_profile = image.get("icc-profile-data") if image.get_typeof("icc-profile-data") != 0 else None

it returns a value for icc_profile. Does this mean the image has a color profile?

[jcupitt]

Ah wait, it's an openslide limitation. There's a profile on your original image, in directory 2:

$ tiffinfo ventana.tif
...
=== TIFF directory 2 ===
TIFF Directory at offset 0x23cc19f2 (600578546)
  Image Width: 51200 Image Length: 99840
  Tile Width: 1280 Tile Length: 1280
  Resolution: 21505.5, 21505.5 pixels/cm
  Bits/Sample: 8
  Compression Scheme: JPEG
  Photometric Interpretation: YCbCr
  YCbCr Subsampling: 2, 2
  Samples/Pixel: 3
  Planar Configuration: single image plane
  Reference Black/White:
     0:     0   255
     1:   128   255
     2:   128   255
  ImageDescription: level=0 mag=20 quality=95
  Software: ScanOutputManager 1.1.0.15854
  DateTime: 2025:09:16 19:53:45
  ICC Profile: <present>, 1811428 bytes

You could try extracting and applying that, it might help.

Are you using a libvips without openslide support? What do you see for:

$ python
Python 3.13.3 (main, Aug 14 2025, 11:53:40) [GCC 14.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import pyvips
>>> image = pyvips.Image.new_from_file("ventana.tif")
>>> image.get("vips-loader")
'openslideload'
>>> 

[jcupitt]

I took the profile out of your original image like this:

$ python
Python 3.13.3 (main, Aug 14 2025, 11:53:40) [GCC 14.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import pyvips
>>> image = pyvips.Image.tiffload("ventana.tif", page=2)
>>> profile = image.get("icc-profile-data")
>>> f = open("profile.icc", "wb")
>>> f.write(profile)
1811428
>>> f.close()
>>> 

And applied it to the image with:

$ vips icc_transform ventana.tif[rgb] x.tif[pyramid,tile,compression=jpeg] srgb --input-profile profile.icc --intent perceptual

I now see (before on the left, after on the right):

Which looks closer to what you wanted.

[Rasaa84]

Thank you so much for your help.
yes, I use pyvips with openslide support. what if I want to use dzsave after transform? is this the code I should use?

import pyvips

image = pyvips.Image.tiffload("ventana.tif", access="sequential")

out = image.icc_transform("srgb",
                          input_profile="profile.icc",
                          intent="perceptual")

out.dzsave("ventana_dzi", tile_size=1024)

[jcupitt]

Don't use tiffload, it's not a TIFF file. Use eg.:

image = pyvips.Image.openslide("ventana.tif", rgb=True)

[Rasaa84]

it worked. Thank you so much.

[Rasaa84]

@jcupitt
Could this solution be dependent on the operating system? It successfully adjusts the color on Windows 11 with Python 3.10, but fails to do so on Amazon Linux 2023 with Python 3.12. I’m using libvips with OpenSlide support in both environments, with the same versions installed:

pyvips == 3.0.0
openslide-python == 1.4.2

No errors or warnings are shown. Are there any other factors that might explain the difference in behavior?

[jcupitt]

You'd need to give a lot more information before I could guess.

I would debug it like any issue: break it into small steps, compare the two systems, discover exactly where they diverge.