well the model already 16 fps
well the model already 16 fps
Fso we can double and make 32 :d
Toh, I didn't realise the measurements are for 80 frames generation (5s at 16 fps). That makes comparsion with Kling even worse as it outputs 150 frames per 4 minute generation
Fye kling is lead 
Fbut in some cases
FWan 2.1 made better output i saw comparisons
Fby the way when i add double it will be 160 frames for us
A@Dr. Furkan Gözükara this is exactly what. I meant on Patreon chat
Ais it even worth the research time when img2 vid exists?
Ji saw people train Lora for hunyuan, i hope this comes to Wan too
Fthere is already Wan training at the library we use
Joh didnt knew that
ABTW multi gpu support for wan which devs are ignoring as you said Dr will make it very competitive
Aimagine 4x A6000 with 10min gen time it would be dope
Fyep and cheap too
Ahow can we push the devs ? maybe you could implement it for us or not possible?
Fi am trying to make kaggle work haha
Fnope i cant
Akaggle?
AI can't imagine if 50min on H100 how much it will be on kaggle 
Ftrying 1.3b
JDoc i just heard Wan can also create audio
Fthey show
Fbut they didnt publish yet
Jok, so we still need MMAudio
Fcurrently yes we have it already also
Ji know, but you know it also needs an update for the 5000er 
Fworking on fps increase thing
Fi can do now
Fgood reminding
A@Dr. Furkan Gözükara can I do good inference with a 700$ gpu?
Alike a 4070
Fit is good as long as fits into vram
Fyou can also get 2nd used rtx 3090
Fshould be even cheaper
Jah okay awesome, thanks! Will check!
Afrom Newegg?
Fdepends on where you live
AGreece
Fsadly this wont work yet
Fno torchaudio for windows for cuda128 
Fbut still testing
Fye no windows. remind me tomorrow
Fwow this is really good compare
FAfps boost ?
Fyes
Fworking on implementation
�@Dr. Furkan Gözükara Can the wan2.1 14b model run on 48 gb vram?
Ait uses 43.9gb
