SixLabors/ImageSharp

Simple conversion of Tiff file to PNG or JPG generates a strange result, even, at certain times, it is in 3D perspective

yuridiniz opened this issue · 6 comments

yuridiniz commented

Prerequisites

  • I have written a descriptive issue title
  • I have verified that I am running the latest version of ImageSharp
  • I have verified if the problem exist in both DEBUG and RELEASE mode
  • I have searched open and closed issues to ensure it has not already been reported

ImageSharp version

3.1.2

Other ImageSharp packages and versions

No

Environment (Operating system, version and so on)

Linux / Windows

.NET Framework version

.NET 6

Description

I have attached the images that are generating this behavior

Steps to Reproduce

O problema ocorreu em uma imagem específica, segue código mínimo para reprodução:

using (var inImg = File.OpenRead("Image_025.tif"))
using (var outImg = File.Create("Image_025.png"))
using (var imagem = Image.Load(inImg))
{
    imagem?.Save(outImg, new PngEncoder());
}

Images

Img.zip

brianpopow commented

The expected decoded image Image_021.tiff should look like this:
expected

It actually looks like this from the main branch:
actual

The tiffinfo is:

TIFFReadDirectory: Warning, Unknown field with tag 347 (0x15b) encountered.
Image_021.tif: JPEG compression support is not configured.
=== TIFF directory 0 ===
TIFF Directory at offset 0x8 (8)
  Image Width: 1666 Image Length: 2359
  Resolution: 200, 200 pixels/inch
  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

So this is a jpeg compressed tiff image. The image decodes correctly, if we set the ycbcrSubSampling to 1, 1 instead of 2, 2, since this is already handled by the jpeg decoding part.
I am not 100% sure if we always can assume ycbcrSubSampling to 1, 1 with tiff compressed jpeg's. @JimBobSquarePants any thoughts?

JimBobSquarePants commented

Does the tag specifically state the subsampling is 2, 2?

https://www.awaresystems.be/imaging/tiff/tifftags/ycbcrsubsampling.html

brianpopow commented

Does the tag specifically state the subsampling is 2, 2?

https://www.awaresystems.be/imaging/tiff/tifftags/ycbcrsubsampling.html

Yes YCbCrSubSampling is [2, 2] here in the example image Image_021.tiff as can be seen in the tiffinfo section above.

JimBobSquarePants commented

I had another dig through our Tiff code (I really do not enjoy how complicated it is to follow) and it appears to me that our jpeg scan decoder is already handling the deinterleaving so we can treat the subsampling as [1, 1] when using the jpeg decoder.

brianpopow commented

I had another dig through our Tiff code (I really do not enjoy how complicated it is to follow) and it appears to me that our jpeg scan decoder is already handling the deinterleaving so we can treat the subsampling as [1, 1] when using the jpeg decoder.

Ok, thanks for confirming. I think what makes it so complicated is the myriads of possible combinations of compressions, pixelformats and colorspaces. When ever I think now we finally covered all possible combinations someone comes up with a new tiff variant.

I will try to make a fix for that. @yuridiniz Is it ok, if we use your testimage in a unit-test?

shapeh commented

..."all possible combinations someone comes up with a new tiff variant."

TIFF stands for Thousands of Incompatible File Formats!

Not helpful but couldn't help myself ;)

  • 😄2