is that 3hrs for the clip...
is that 3hrs for the clip...
Fye 
need more vram until authors fix
need more vram until authors fixFtesting video to video
C
finally subscribed. keep up the good work.Flets go video to video
Fawesome and welcome
Fit worked but so small changes will try higher denoise
FFalmost ready . only need to test batch image to video
Xfor image to video does it use the same resolution for the 1.3b model?
Fno it is 720p
F14b model
Fthus it need more vram
Fall right all ready time to test on 80 GB A100 and make kaggle notebook
Ftesting on massed compute A100
Ffull model uses 52.16 gb - 14b image to video
1Is that the FP32?
Fbf16
F
Fbut this is almost as if fp32 vram
Fwe need 1.3b image to video model :d
Fi will test on h200 lets see the speed
Zall very exciting
Fye really cool
Fi will publish on Patreon hopefully today before sleep
Ffinal test
F14b text to video broken atm though i expect fix tomorrow
Fdownload speed with special tech
Fh200 is fast
Fh200 is 1.84 second id on my rtx 3090 ti around 8.5 second / it - 1.3b model
Ftext to video 14b fits into 24gb vram - min 17 gb
Fh200 is 36 second / it for 720x1280 image to video at 14b for 81 frames
Fwaiting first image to video
Flol i worked full day today at least 14 hours now
Fjust wow
F
YOH
Yyou have to tell me how to do this specific vid from prompt to approach
Yso gooood
Yokay imma try it lol
Ypls tell me you already dropped a tutorial in your patreon
Clooks like he did
Ajust amazing
Aalso refreshed my Patreon subscription
ADoctor I think it is very near kling tbh
AI'm going to try today and experiment with it
Tso around 30 minutes on h200 per generation? That's nuts
Thow many iterations per frame does it do?
A@Dr. Furkan Gözükara what gpu should I rent for the large 14b model on massed ?
Dgood grief, how big are these models - 60+gb to download
