Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
8th April 2024, 19:54 | #1623 | Link | |
Registered User
Join Date: Oct 2018
Location: Germany
Posts: 1,002
|
Quote:
Unfortunately, I cannot reproduce the problem on my PC. See the video. However, I am using Win10. Have a look here, there have been problems with other functions that were fixed with Avisybth 3.7.3post test 14. Your description of the problem is exactly the same, namely when releasing the clip and reinitializing. What Avisynth version are you using exactly? I have Avisynth 3.7.3post test 14 on my PC and have no problems with your script. If you have a different version then please test the latest Avisynth test version. Let me know in any case.
__________________
Live and let live |
|
8th April 2024, 20:03 | #1624 | Link | |
Registered User
Join Date: Oct 2018
Location: Germany
Posts: 1,002
|
Quote:
Code:
Frame %F / %FC %TR - (%T) %BM %POS %YUV %RGB \t\t %Z %Wx%H (%AR) - %FR fps - %FB %PS - %CS (%BITbit) (Audio %AUBbit, %AURHz) %DYR
__________________
Live and let live |
|
17th April 2024, 15:17 | #1625 | Link |
Registered User
Join Date: Sep 2007
Posts: 5,377
|
-1 float read issue with color picker
blankclip(pixel_type="RGBPS", colors=[-1,-1,-1]) reads as "0" in the color picker ; the true rgb value "disappears" , only the 8bit rgb value is displayed
blankclip(pixel_type="RGBPS", colors=[-2,-2,-2]) reads as "-2" Not an avs+ core issue because blankclip(pixel_type="RGBPS", colors=[-1,-1,-1]) avs script to ffmpeg correctly uses gbrpf32le for avs demuxer , writing out exr and reading with vapoursynth reports the correct -1 float pattern tiff if you want to test. The values for the 5 patches should be RGB -2.-1,0,1,2 https://www.mediafire.com/file/bu757...ttern.zip/file BestSource2 can load float files into AVS+ now, based on those observations, it's not a plugin issue either EDIT: Interestingly, if any of the values are not -1, all 3 values now read correctly . Problem only occurs when all three are -1 e.g colors=[-1,-1.01,-1] now reads all the values correctly Last edited by poisondeathray; 17th April 2024 at 16:35. |
18th April 2024, 21:18 | #1626 | Link | |
Registered User
Join Date: Oct 2018
Location: Germany
Posts: 1,002
|
Quote:
2.7.7.3 Pre-Release
__________________
Live and let live Last edited by gispos; 18th April 2024 at 22:31. |
|
18th April 2024, 22:53 | #1627 | Link | |
Registered User
Join Date: Sep 2007
Posts: 5,377
|
Not sure if there is "valid range" . Normalized range is [0,1] .
Usable range is much greater than [0,1]. Large positive and small negative values exist for 32bit float in actual usage,actual files in other programs. I've seen .hdr's , .exr's that have usable highlight details in the +50 range, usable shadow details as low as -2 . Quote:
|
|
Yesterday, 20:48 | #1630 | Link | |
Registered User
Join Date: Jul 2018
Posts: 1,075
|
Quote:
|
|
Yesterday, 21:54 | #1631 | Link | |
Registered User
Join Date: Oct 2018
Location: Germany
Posts: 1,002
|
Quote:
Blacker than black... deadlier than dead
__________________
Live and let live |
|
Yesterday, 23:06 | #1632 | Link |
Registered User
Join Date: Jul 2018
Posts: 1,075
|
It may be simply digital undershoots from oversharpening or very 'video-look/makeup'. With new HDR with very high above 1.0f max white the undershoots at sharp transients to black may go much below 0.0f of nominal black.
Some still very few ringing samples sequences may have undershoots about -0.3 of nominal white to black range With not very big (for HDR) white or 3.0f..3.5f the most negative sample value may be close to -1.0f. So good safe footroom for undershoots is about -30..-40% of max white is expected in perspective systems. The typical 8bit narrow range encoding have only about 15/219 ~= -7% footroom only for undershoots. With float encoding systems it is possible to have a bit more sharper transients with still low ringing because footroom is unlimited in comparison with integer systems with only about -7% footroom at any bitdepth. Last edited by DTL; Yesterday at 23:33. |
|
|