Jim Leyden

 and I'm still running into issues in terms of the "seam" of the loop. (I'm running on windows) It's more prominent with a high rate (which is my main concern, as the loops need to go faster in different segments of the program). At some of the seams it "skips," and at others it sort of seems to "drag" as it approaches the seam. What also concerns me is that it seems to lock up other things that are happening purely on the GPU when it's placed within the project and not just a testing patch. I'm not seeing a drop in framerate when this occurs, however.

As it stands I'm wondering if a codec change will help me, but I've already tried the larger, more lossless/non-temporal codec options available to me (no pro-res as I'm on windows) and it seems worse, to be honest. The lock-ups are longer (and I get weird skips to the beginning).

I'm not really sure where to go from here. I'm loading into RAM, the cache_size is big enough, set to uncompressed frame size and it loads the whole clip. Everything is in h.264, the best performance I can get is from that codec, keyframes in the encode are set to keyframe every 1 frame, performance on the hardware is a high percentage but definitely not maxing out (cpu does spike when the seam comes though - would that be an issue?).

How is it that I can get nice, smooth playback at 1920x1080 until this seam? Is looping video really that taxing? I know the software probably work in completely different ways but when I play the loop through in After Effects there's no stutter whatsoever. I just wish I could find a good solution, this is really putting a huge roadblock in front of what I'm working on.

I ran through a lot of the solutions listed here and I'm still running into issues in terms of the "seam" of the loop. (I'm running on windows) It's more prominent with a high rate (which is my main concern, as the loops need to go faster in different segments of the program). At some of the seams it "skips," and at others it sort of seems to "drag" as it approaches the seam. What also concerns me is that it seems to lock up other things that are happening purely on the GPU when it's placed within the project and not just a testing patch. I'm not seeing a drop in framerate when this occurs, however.

As it stands I'm wondering if a codec change will help me, but I've already tried the larger, more lossless/non-temporal codec options available to me (no pro-res as I'm on windows) and it seems worse, to be honest. The lock-ups are longer (and I get weird skips to the beginning).

I'm not really sure where to go from here. I'm loading into RAM, the cache_size is big enough, set to uncompressed frame size and it loads the whole clip. Everything is in h.264, the best performance I can get is from that codec, keyframes in the encode are set to keyframe every 1 frame, performance on the hardware is a high percentage but definitely not maxing out (cpu does spike when the seam comes though - would that be an issue?).

I ran through a lot of the solutions listed here and I'm still running into issues in terms of the "seam" of the loop. (I'm running on windo

questions-on-encoding-playback-video-for-max-rate-looping-etc···

you can post your patch and I'll take a look to see if there's anything obvious to be done. looping stutters are pretty annoying I know. i've tried to offer solutions to avoid them, with the most effective being loadram ranges, allowing you to load only the start and end points into memory.

but there's probably still work to be done to make this better for all platforms and video engines.

also want to mention, prores codec is supported by viddll so you can use it on windows. prores422 would be the first place i'd start for non-temporal codec testing.

but there's probably still work to be done to make this better for all platforms and video engines. 

also want to mention, prores codec is supported by viddll so you can use it on windows.  prores422 would be the first place i'd start for non-temporal codec testing.

Perhaps, when using large, and generally - all sizes of the loadram ranges, it is necessary to forbid unload the start and end points from memory ?

@rob I'd show you a patch but it'd look awfully familiar as everything I've been testing with has been something you posted on here somewhere. How does loadram "range" work? As it stands I'm able to load the entire loop into RAM and still get the skips. Is loading just the beginning and end somehow better? How is that done?

Also, I did notice prores422 is supported by viddll but I can't export any of my loops to it with the Windows version of After Effects I did however try photojpeg and animation codecs for .movs, and those didn't help and in some cases made it worse.

jit.movie has just generally been frustrating as I know my hardware is decent but I can't get movie/movie files directly onto a GPU, and the CPU is bottlenecking the shit out of everything I'm trying, even beyond these loops.

@rob I'd show you a patch but it'd look awfully familiar as everything I've been testing with has been something you posted on here somewhere. How does loadram "range" work? As it stands I'm able to load the entire loop into RAM and still get the skips. Is loading just the beginning and end somehow better? How is that done?

Also, I did notice prores422 is supported by viddll but I can't export any of my loops to it with the Windows version of After Effects I did however try photojpeg and animation codecs for .movs, and those didn't help and in some cases made it worse.

Questions on Encoding/ Playback Video for Max - Rate, Looping, Etc...