Missing or Incorrect contrastChecks and sizeChecks Video/Image Outputs

[scherich-sqex]

Hi,

I'm encountering issues when processing certain videos in Fonttik. While some input
videos correctly generate contrastChecks+sizeChecks videos and images along with
the corresponding text logs, others only produce the contrastChecks+sizeChecks videos
without any textboxes highlighting the detected text. In these cases, the generated
videos appear to be exact copies of the input videos, and no output images are generated.

The application does not display any errors or warnings, so it's unclear what might
be causing the issue.

Since the process works as expected for other videos, I'm unsure whether this is
due to a configuration setting or something specific to the input files.

Below is a snippet from the text log of a video that failed to generate the expected
outputs:

Processing video frame 78 - 14.2%
Word at (1725, 896) doesn't comply with minimum height 28, detected height: 23
Processing video frame 84 - 15.2%
Word at (1725, 896) doesn't comply with minimum height 28, detected height: 23
Processing video frame 90 - 16.3%
Word at (1725, 896) doesn't comply with minimum height 28, detected height: 23
Processing video frame 96 - 17.4%
Word at (1725, 896) doesn't comply with minimum height 28, detected height: 23
Processing video frame 102 - 18.4%
Word: [104 x 42 from (699, 104)]  doesn't comply with minimum luminance contrast 4.5, detected contrast ratio is 4.432460482247436 at: [104 x 42 from (699, 104)]
Processing video frame 108 - 19.5%
Word: [102 x 42 from (701, 104)]  doesn't comply with minimum luminance contrast 4.5, detected contrast ratio is 4.414430780041194 at: [102 x 42 from (701, 104)]
Processing video frame 114 - 20.6%
Word: [104 x 42 from (699, 104)]  doesn't comply with minimum luminance contrast 4.5, detected contrast ratio is 4.439494602569445 at: [104 x 42 from (699, 104)]
Processing video frame 120 - 21.6%
Word: [104 x 42 from (699, 104)]  doesn't comply with minimum luminance contrast 4.5, detected contrast ratio is 4.44253843772044 at: [104 x 42 from (699, 104)]
Processing video frame 126 - 22.7%
Word: [104 x 42 from (699, 104)]  doesn't comply with minimum luminance contrast 4.5, detected contrast ratio is 4.439406085581356 at: [104 x 42 from (699, 104)]

The end result of the video is:

SIZE CHECK RESULT: FAIL
CONTRAST CHECK RESULT: FAIL
SIZE CHECK RESULT: FAIL
CONTRAST CHECK RESULT: FAIL

Below is a comparison between a video that exhibits the issue and one that processes
correctly, along with the system specifications used for testing:

System Info

• Operating System : Windows 11 Enterprise
• CPU : 13th Gen Intel(R) Core(TM) i9-13900KF
• Memory : 64 GB
• Video Card : NVIDIA GeForce RTX 4080

Video Info

Property	Video with issues	Video without issues
File Type	mp4	mp4
File Size	7.75MB	40.53MB
Format	yuvj420p	yuvj420p
Resolution	1920x1080	1920x1080
Aspect Ratio	16:9	16:9
Duration	9.38 seconds	64.25 seconds
Color Depth	8-bit	8-bit
Bit Rate	6,758,030	5,121,277
Frame Count	563	3855
Frame Rate	60	60
Codec	h264	h264

Video with issue: https://github.com/user-attachments/assets/606a3866-b2e3-4921-87c8-2dd820a1b049

Any insights into what might be causing this discrepancy would be greatly appreciated.

Thank you!

[RicardoSocas]

Hello!
We've added a new release with the latest changes. Please try it out to see if your videos get better results now.
We also recommend analysis on images instead of videos so you know exactly what the tool is using for it's analysis.

[scherich-sqex]

Hi @RicardoSocas,

Thank you for the reply and updated build(1.0.2)!

I have tested it out with the video I attached earlier and the issue still seems to be present.

Additionally, another issue I encountered in the previous build that is still present is that processing seems to get stuck for some videos:
https://github.com/user-attachments/assets/2f34158b-3b33-49c7-b42e-da27cf206a23 .
That is the tool gets stuck at a certain processing percent even after leaving it running for hours.

Regarding the use of the tool for images, it is indeed my preferred option, but there are cases where videos might need to be processed.