PIN2DMD Layered ColorMask Misbehaviour

[sudsy7]

Here's one that I don't understand (and probably @lucky1 can quickly assess what I'm doing wrong here).  There is a scene in Breakshot (consisting of about 50 frames) that is used both in attract mode and for the end-of-game match sequence.  Both animations use the same frames, but of course the match sequence has numbers overlayed on it.  I know the LCM is usually used on video modes, but I thought it would be perfect for this application as well, and it would take care of both the attract and match animations all in one fell swoop.  In fact, I couldn't think of any other mode that could actually do this particular scene better.  So I created the LCM scene, colored and masked all the match numbers and ran it on my pinball machine.   It works perfectly in the attract mode animation (see first video).  But it only "sort of works" in the match animation (see second video).  It's difficult to see in the video (looks worse in person), but it appears like the frames are intermittently being colored, so it looks "choppy".  If I isolate a frame (see picture), it appears what I'm intermittently seeing is white, black and 2 shades of grey (and probably not by coincidence, those are the 1st four colors in the color palette for this scene).  I'm using 64-color mode, latest editor, and f/w 4.20.   The pink/red match colors remain solid throughout the entire animation, however.  Any ideas of what the problem might be and/or suggestions how to fix?

 

[lucky1]

Videos say nothing about what is done wrong. Please upload the project files and send me the link.

[lucky1]

I had a look at your project and generally everything looks fine in it.

The reason why this is happening seems that the numbers get recognized correctly but the

detection of the background scene seems to fail in transitions which results in only the numbers appearing in colors

and the rest of the screen in the default colors which is black and white. This problem is Capcom specific. Maybe a RAW dump of thoses scenes can help finding a better suitable trigger/hash for those frames.

 

[sudsy7]

OK, that's interesting.  It seems odd that it reliably detects what are essentially the same background frames (same masked hashes) in attract mode, but not during the match sequence.  I guess it really is still detecting all of them in the match sequence, but the intermediate transition frames (that are not visible to the naked eye when using a only a 4-color colormask) are messing things up.  I still wonder why those transition frames are reverting to the default black & white colors - I thought, by design, a colormask frame would persist until a new hash is recognized, and these transition frame's hashes are not set.   Anyways, I will get some raw dumps on it and see if it can be improved.  Thanks for taking a look at it for me.

[Terranigma]

Hey Sudsy,

 

Your colourisation looks great. Do you have any intention of releasing this once finished? 

 

Thanks

[sudsy7]

Thanks, I sure do, but I have a long way to go, and I can't release it with "bugs" like this.  The LCM, Replace & Colormask Sequence modes, and 64-color palettes have made handling difficult scenes like this one possible.  Lucky1 has done a great job continuously improving PIN2DMD firmware and editor. 

Edited September 7, 2021 by sudsy7

[lucky1]

13 minutes ago, sudsy7 said:

I still wonder why those transition frames are reverting to the default black & white colors - I thought, by design, a colormask frame would persist until a new hash is recognized, and these transition frame's hashes are not set.

 

This is the case when no hash is detected in the new frame, but in your case there is a detection of the number part which clears the previous mask.

[Terranigma]

10 minutes ago, sudsy7 said:

Thanks, I sure do, but I have a long way to go, and I can't release it with "bugs" like this.  The LCM, Replace & Colormask Sequence modes, and 64-color palettes have made handling difficult scenes like this one possible.  Lucky1 has done a great job continuously improving PIN2DMD firmware and editor. 

 

That sounds great, thank you. I will add your project to the main list as a work in progress. Good luck with your project!

[sudsy7]

19 minutes ago, lucky1 said:

This is the case when no hash is detected in the new frame, but in your case there is a detection of the number part which clears the previous mask.

I see now - that makes sense.