for the sort by similarity app, what kind of speeds should i be expecting? right now it takes about
for the sort by similarity app, what kind of speeds should i be expecting? right now it takes about an hour to process 20 images
Fit is wrong
Fit means not using your gpu
Fit uses tensorflow so it is extremely delicate to the cuda and cudnn libraries
Fin your system
Fit sucks from this perspective actually
Dyeah its all on the cpu atm. not sure why
Di have all the prereqs installed. i think it might be defualting to a cached version of keras and protobuf that aren't compatible
Dthank you. might be time for a clean install again lol
Fit is related to cuda and cudnn 99%
Fi plan to move cuda 12.8 but torchvision and torchaudio cu128 still not ready. after that hopefully every app can work on 5000 series as well
Fwow this week was so busy still reading all emails 
Fi dont remember mersin was this cold ever
F3 celcius outside freezing here haha
Ftime to add SkyReels to our swarmui downloader and test
Fvery dynamic first test
F
Pyou have lost of identity
Nhow to make a local server through pinokio? to share like a browser to a neighboring computer?
Fyep
Fi am still searching best params
Fno idea for pinokio
Fwhich app?
Ffixed metadata of models. gonna upload all for you
F
Ranyone try the skyreels video model?
Fyes i did
Fpreparing tutorial
Fpurchased daily higher upload internet to upload models faster
Aon a scale of 1-10 how close is it to Kling ?
Abased on your experimentation
Fi dont experiment with kling so cant tell
Aok is the tutorial dropping today?
Fif i can finish
TBought a Pro subscription for Kling for testing.
The 1.6 pro mode is impressive, but have limitations. Much better for cinematic (realism) videos than cartoon animation style (decent with 3D pixar style though).
Getting something specific what you want requires prompt engineering trial and error and it's quite expensive considering each generation costs 0.3-0.5$ each and you may need 10 tries to get specific result.
Overall quality is very impressive
The 1.6 pro mode is impressive, but have limitations. Much better for cinematic (realism) videos than cartoon animation style (decent with 3D pixar style though).
Getting something specific what you want requires prompt engineering trial and error and it's quite expensive considering each generation costs 0.3-0.5$ each and you may need 10 tries to get specific result.
Overall quality is very impressive
Acan confirm been a pro user for 2 months
Aalso @Timson , the daily credits will continue for non-stop subscribers before the 22th
Tyeah, I know, but I subbed exactly 22th, unlucky
TI think they are going to reduce daily credits amount though, since it is not fixed amount
Aeither way pika is catching up slowly so competition will be beneficial for us
Fgoogle veo also started but extremely expensive as far as i know
F5 second is 1.25 usd on fal
Fi plan to test hunyuan on a100
Fit is 1 hour 1 $
P@Dr. Furkan Gözükara i tried everything to improve my flux dev+dreambooth finetuning outputs but the issue is the model is not stable sometimes it gives good , sometimes bad, can't give to my client in this conditions.I has used 250 images of the same person, what would you suggest will I need 1000 photos to get the most realistic images without any errors?
Tthere are rumors of official hunyuan img2vid in feb-march (announced in their discord my staff member in unofficial conversation as his estimate)
F250 images is good number. if not working accurately try to improve dataset
Tto me img2vid is millon times more interesting as it gives so much control over the result
Ttext2vid is just too random results
