logo Sign In

nightstalkerpoet

User Group
Members
Join date
25-Oct-2010
Last activity
30-Jun-2025
Posts
1,108

Post History

Post
#1021489
Topic
STAR WARS: EP V &quot;REVISITED EDITION&quot;<strong>ADYWAN</strong> - <strong>12GB 1080p MP4 VERSION AVAILABLE NOW</strong>
Time

I now just really want to see Ady combine TPM/AotC into a single Episode 1, Fix RotS into a less revealing Episode 2, and streamline Rogue One just a bit (add crawl, remove background fanservice) into a perfect gritty Episode 3. Knowing I could finally watch a Star Wars marathon without cringing would be amazing 😄

Post
#1015826
Topic
Info: The Ultimate Super Resolution Technique
Time

One of the things I’ve noticed about about any type of noise reduction or temporal super resolution is a slight (or occasionally intense) flattening of the image. I was curious, is there a way to generate a depth map from the original video, upscale the map and reapply it after all filtering and upscaling has taken place? More than anything, depth is the one feature that I’ve found Super Resolution techniques to be lacking in. This doesn’t apply to the stacking concept really, since it is single frame temporal.

Post
#1015223
Topic
Star Wars saga - Extra Extended Edition (1080p) (* unfinished project / WIP *)
Time

Actually, what about combining the methods? Using automated color tools to add color information into the picture, then using color match to adjust those colors?

Perhaps someone familiar with Matlab could help with this? Looks like it could be pretty amazing for establishing a base image to match from. http://www.cs.huji.ac.il/~yweiss/Colorization/

Post
#1013663
Topic
Feedback Wanted: my ColorMatch LUT Sharing
Time

Thanks 😃

Yes the highlights are a bit bright, as are the darks. I keep tinkering with it but it’s difficult to find a solution that gives each scene the right color and brightness.

I’m finding that the best shots to use for developing a color match model for Dre’s tool are wide shots of the falcon cockpit. On the wall behind them is a display showing red blue and green. One of the shifters is partially yellow. With those colors so isolated it’s easier to find the right values. Star Wars is the easiest because with Luke, Obi-Wan and Han all having similar yet different color clothing, you can get a better feel for the subtleties. The issue with these shots is the bluray is so dark that when the color is fixed the lights are really bright and it throws off brightness and contrast.

I also tend to use gimp to adjust the initial match image to fix the colors that are still wrong in the rest of the film, even if the small adjustments aren’t very visible in the match shot.

Post
#1013494
Topic
WOWOW Star Wars Saga (Released)
Time

I’ve seen a few tests where MPEG2 is better at maintaining all details across the picture, while AVC/H.264 better maintains detail that the human eye is likely to see. That is why smaller things like shadow detail and potentially background items that are not the main focus may be better preserved. While you’d imagine that these smaller details aren’t as necessary, they add an extra level of depth and are very useful if you need to upscale, since the extra detail in those areas provide better approximations for what the image should look like at a larger size.

http://www.streaminglearningcenter.com/articles/h264-vs-mpeg-2-quality.html shows what I mean - look at the buildings in the background and the people in the crowd. That website does not show them at the same bitrate, but you can still see how the bitrate distribution differs.

Post
#1013314
Topic
Avisynth Scripter/Programmer wanted for Mode Average Plugin
Time

Just a concept for a mode average plugin - I’ve always thought this would provide better results than a median average, but have never been able to find one and am unsure how to write avisynth plugins - I’m willing to learn but it would take me a while and I think this would be worth someone with more experience compiling quickly.

So, conceptually you’d have a 5 capture average (though with this setup, you would be able to use up to 13 captures. In the case of all different pixel values, even numbers of captures would find the mean of the middle two values, while odd numbers would find the median.)

Finding the mode of each pixel value, A(Pixel Value) corresponds with N(Number of clips with that pixel value), B corresponds with O, C with P, D with Q, E with R.

Initial values are N=1, O=0, P=0, Q=0, R=0 (S,T,U,V,W,X,Y,Z all = 0)

If B=A, N=N+1. If B doesn’t equal A, O=O+1. Check if C exists. If so, go to next line. If not, check if N=O(letter). If so, find the mean of A and B. Otherwise, go to next function.

If C=A, N=N+1. If not, check if C=B. If C=B, O=O+1. If C doesn’t equal B, P=P+1. Check if D exists. If so, go to next line. If not, check if N=O=P. If so, find the median of A,B, and C. Otherwise, go to next function.

If D=A, N=N+1. If not, check if D=B. If D=B, O=O+1. If not, check if D=C. If D=C, P=P+1. If not, Q=Q+1. Check if E exists. If so, go to next line. If not, check if N=O=P=Q. If so, remove the highest and lowest of A-D, and find the mean of the two middle values. Otherwise, go to the next function.

If E=A, N=N+1. If not, check if E=B. If E=B, O=O+1. If not, check if E=C. If E=C, P=P+1. If not, check if E=D. If E=D, Q=Q+1. If not, R=R+1. Check if F exists. If so, go to the next line. If not, check if N=O=P=Q=R. If so, find the median of A-E. Otherwise, go to next function.

SECOND FUNCTION:
Find highest value of N-Z. Select the corresponding pixel value A-M, and report that as the selected pixel value.

Next pixel.

Any thoughts or interest?

Post
#1012672
Topic
Feedback Wanted: my ColorMatch LUT Sharing
Time

I have been working on a “one pass fixes most” for ANH on Bluray and thought it might be nice to start a thread for everyone to share LUTs and ColorMatch Models and get feedback.

Here’s my initial one for the EP4 Bluray - attempting to bring it closer to theatrical color timing: https://drive.google.com/file/d/0By9sWLF9s-s6ajA2bnNNUW4wZkk/view?usp=sharing

Would love feedback.