I plan to compare GGUF vs our APP tomorrow hopefully
I plan to compare GGUF vs our APP tomorrow hopefully
Zwhenever you get a chance Dr, let me know how to speed up the generation time in MC, thanks!
Hopensource video models still have long way to go
Hthink about sd1.4 first release
Hand we got sd1.5 fine-tunes in just a year
Hquality improvements are just incredible between sd1.4 and sd1.5 fine-tunes
Hwe gonna see better, faster and local video models by end of 2025, im totally sure with that
AThis one I made is slightly better restoration, but not by much!
TFound it on X. How true not sure, but need more investigation.
JRegarding upscaling in swarmui, what is the 4xrealwebphoto model that you use in tips-info @Dr. Furkan Gözükara ? Or which one is the best to use (for realistic portrait photos)
Hinterestingly enough, training a ~250 image dataset on wan2.1 14b "only" takes 27gb gpu at res 1024 and rank 128, I wonder how far it can be pushed with a 80gb gpu
Hbut to be fair it's quite slow per iteration (~23sec)
Hneed to try videos next but I really want to push that training to the max
HBUT the lora is 1gb damn
Fthis guy doesnt know
Fhe confused telemetry data 
Fit is here
FOpenModelDB
OpenModelDB is a community driven database of AI Upscaling models. We aim to provide a better way to find and compare models than existing sources.
Fmy unified downloader also downloads it
Fhow good it works?
Fi plan make a gui for this
Fby the way sd 3.5 large training getting improved. now we gonna have trained clip l and clip g into embedded file
HI'm still training, it's very slow but it works well on a A6000, I've not trained videos yet but I will work on it this week end, but it's promising tbh, the lora file is 1gb tho, I'm waiting for a few epoch and will test and let you know
F1gb is not big
HFor a lora ? Would you recommend going rank 256 ? I don't really care about the size or the time it takes to train tbh, as long as the results are there
F128 usually great
DIs this WAN?
AFreally good
TAre you planning to research gguf for flux dev as well?
Fyou mean compare gguf of flux vs like fp8 or bf16?
Tyep, and also a tutorial on how to use it in swarmUI
Fi plan to update my downloader and seperate
Fflux hunyuan wan etc and including gguf as well
Falso found a solution for kaggle
Ftesting now
Fwan 2.1 1.3b model should work fast
Tawesome tutorial as always, great work!
TMy only suggestion is using 30 fps video examples, since it's easier to compare with Kling (it always generates 30fps)
Fi think i will add rifle to increase fps to 30
Fi am gonna compile flash attention for swarmui - python 3.12.8
TTo me bottom line here is - Wan 2.1 14B is not practically usable.
It is on par with Kling 1.6 Pro mode in terms of fidelity, but Kling is both cheaper and much much faster.
50 minutes per generation is not currently usable in AI creative process that requires lots of retries and prompt iteration for any meaningful results.
L40S is at best 1.2$/h currently so it's 1$ per generation (not including time and effort for initial setup) compared with 0.5$ per generation at Kling that completes in 4 minutes.
It is on par with Kling 1.6 Pro mode in terms of fidelity, but Kling is both cheaper and much much faster.
50 minutes per generation is not currently usable in AI creative process that requires lots of retries and prompt iteration for any meaningful results.
L40S is at best 1.2$/h currently so it's 1$ per generation (not including time and effort for initial setup) compared with 0.5$ per generation at Kling that completes in 4 minutes.
Ftrue but difference may come at fine tuning training
Fwhich i plan to cover
Jtrue but i prefer local, unrestricted, uncensored creations 
Fi improved prompt enhance
TTrue, fine tuning and uncensoring are big advantages, but 50 minutes is too much still
Fnow will use like 11 gb
Fabout to publish update
Falso compiling flash attention for improving our swarmui rtx 5000 series upgrade
Fpython 3.12