[Mr.Foxx] Commissions - Requests - Pinups - Short Stories

By MR.FOXX
Blackstone+Request+November+2024.gif
 
Professor Sinister said:
I find Animated PNGs make a much better output than Animated GIF since they do support TrueColor. But the file size can be prohibitively huge so at that point, AV1 is the way to go
Click to expand...

I agree. Alas Animated PNG's are not part of an official standard and instead are a hack, so it's support has to be included in image viewers and web browsers to work.

Add to that, gif files are paletted (8bit/256 colors), while PNG the trucolor will be 24bit, so likely at least 3x larger, with no guarantees the lossless compression can find good patterns. WebM and AV1 are both preferred.

Though years ago i has a program that separated out colors in an image and made a trucolor gif, but that was by adding 255 colors on every frame until the image was completed, making the effect of it being painted. Looks pretty cool... i should probably offer it as a tool...

Though i did find a BadDragon from a 2015 promotion... But as it really doesn't belong here i'm hesitant to post it.
 
Ryze said:
By MR.FOXX View attachment 1965359
Click to expand...

I know everyone has their own tastes, but I notice a certain preference from the author for this character, and honestly, I can't understand it. In my opinion, the characters from the stories "Chore Day," "Forbidden Love," "Hypnosis," and others, are prettier and more desirable than this character. But... everyone has their own tastes, right?! All we can do is respect that and hope to see these characters again someday.
 
Kretua said:
Yes, it's perfect for that. To share the load. But a 24GB model won't fit into two 12GB cards because it only handles VRAM separately. Yes, there are multi-GPU nodes. But you've made me curious, I'll look into it.
Click to expand...
Extremely late reply but might be applicable to you, seems like torch compile (Triton optimisation ) is bugged on pytorch 2.9, it results in massive vram usage when compiling and generally workload hanging or crashing. A lot of workflows come with this and might be a reason why you're struggling sometimes. I bypassed this and was able to do 720p videos in I2V without block swap offloading which wasn't possible before
 
DrSlobb said:
Extremely late reply but might be applicable to you, seems like torch compile (Triton optimisation ) is bugged on pytorch 2.9, it results in massive vram usage when compiling and generally workload hanging or crashing. A lot of workflows come with this and might be a reason why you're struggling sometimes. I bypassed this and was able to do 720p videos in I2V without block swap offloading which wasn't possible before
Click to expand...
Thanks for the info, I'm currently using PyTorch 2.7. The context was about using two GPUs as an option.
 
llermito said:
Fantastic, my friend! The details of the sofa fabric and how the knees press against it are superb. I know a lot of people will say I'm an idiot, but besides faces and bodies, I like to focus on the details that make all the difference in the quality of a piece.
I'm your number one fan!;)(y)
Click to expand...
i like you.....

first to real z-image
to video wan2.2
 
You must log in or register to reply here.
Back
Top