Which version is truly the best in your experience

#216
by jdeepshi - opened

Hi @all . In your experience, which version has produced the best results for you for a variety of prompt( especially nsfw)? Thanks in advance.

I still using V10.
I want to try Mega, but I cannot use the I2V function even through I use phr00t's workdflow and setting.

I'd like to see a screenshot of your workflow when I2V doesn't work for you.

I only use Mega v12.

I'd like to see a screenshot of your workflow when I2V doesn't work for you.

I only use Mega v12.

I use the workflow in mega_V3 folder, I did not change any setting but just delete the End Frame (Optional).
This is good if you can teach me how to use.
image

I'd like to see a screenshot of your workflow when I2V doesn't work for you.

I only use Mega v12.

I use the workflow in mega_V3 folder, I did not change any setting but just delete the End Frame (Optional).
This is good if you can teach me how to use.
image

try this one first: https://civitai.com/models/2143777/caveman-easy-wan-22-aio-nsfwsfw-rtx-3060-12gb-vram-gguf-workflow
the original workflows for ggus also broken for me

i have a continuous workflow with overlap and stuff that works great for the "old" models, but i haven't figured out how to do that with the mega models yet. However from my single tests i wasn't blown away by either of them. The abilities are obviously great but the videos itself weren't. Mega 12 for example made extremely static videos. And i found remarkably few occasions where a start and end frame would be useful for me.

My favorite is the (non-mega) V7. While it's image quality seems weirdly bad especially at lower resolutions, it is by FAR the best (i tried many but not all) for anything with fast action and physical reactions built in. No other later model was as good at that. I wonder what Lora mix was responsible for the physicality of V7. Arguably for less physical actiony videos it may be too ott. And there is no nsfw version (in case one thinks of that physicality).

i have a continuous workflow with overlap and stuff that works great for the "old" models, but i haven't figured out how to do that with the mega models yet. However from my single tests i wasn't blown away by either of them. The abilities are obviously great but the videos itself weren't. Mega 12 for example made extremely static videos. And i found remarkably few occasions where a start and end frame would be useful for me.

My favorite is the (non-mega) V7. While it's image quality seems weirdly bad especially at lower resolutions, it is by FAR the best (i tried many but not all) for anything with fast action and physical reactions built in. No other later model was as good at that. I wonder what Lora mix was responsible for the physicality of V7. Arguably for less physical actiony videos it may be too ott. And there is no nsfw version (in case one thinks of that physicality).

V10 is worse, but there is a NSFW version, not in this channel tho. V10 needs much more handholding and spoonfed about each pixels in the scene, unlike V7, but interpret much more concepts whatsoever

Sign up or log in to comment