Seeing alot of new ltxv material being posted over the past week or two. I've never used it before but this workflow seems like just the thing to get me to try it. ^_^

An interesting workflow, removed the ai part to replace it with Griptape which is the ai node i use.
the vrm memory reserve (god I hope thats not malware) is awesome and prevented my os from closing comfyui to protect my ram lol (allowing me to up the rate to 20 seconds on my 3090)

Sadly the ltx merged model still has many flaws with how genitals meet... We so need base models that are fully trained, and not trained on poisoned data...this age of censorship is a nightmare.... all these cool new toys but if you want to make anything "naughty" then oh no... you cant do that...

I would recommend a note on the image section to specify the 0 setting to remove the end image, I know its in your vid but its easy to miss.

Wow this is great! So its the official model but with loras baked into it? Similar to wan 2.2 nsfw models?

why in my video the character says what's in the prompt? How can I describe the scene without the perso to say the prompt?

Hi, I'm trying to do image to video. I watched your youtube video guide. The image is the 1st frame of the video, and then the rest of it has nothing to do with the image. Almost like it's doing a text to video for the rest of it.

This workflow is really great. It is very straight to the point using the only necessary nodes optimized and balanced for great quick generations.

Good workflow but if I try to do i2v at 1080p it creates very cursed things lol. Default res is good but low quality. I don't get cursed things on the default i2v workflow from LTX2. Thoughts on what it may be?

When trying higher resolutions I get so weird results! Like mouths for nipples, sometimes breasts all over the legs and body etc. Very peculiar! Any idea if it is some setting or if it is a limitation with the nsfw model?

"LTXVNormalizingSampler

The size of tensor a (33264) must match the size of tensor b (4290048) at non-singleton dimension 2"

I have this error(((

Where can i change the scheduler?

Tried bypassing the last frame node. Is it really needed? I get error with it.

And why do the characters have bounce-like movement?

My friend, your workflow is excellent. I couldn't find the video on YTDBE. Even though the workflow is NSFW, the models encourage people to develop in different areas. I'm doing non-NSFW projects with your workflow, and it contributes to my development. We used to watch your training video occasionally to get ideas on different topics, but the video has been removed. It's really frustrating to see such hard work disregarded. Keep up the good work, because you're succeeding in this somehow. Thanks.

Excellent Workflow ! takes 10 Seconds to Generate with LCM. at 720p. Testing other parameters with Loras

Don't waste your time. It has error after error and it never ends installing nodes. IDK if the creator made an update and forgot something or what. But this doesn't work and for some reason, freezes ComfyUI all the time.

Edit: Dude... This is stupid. There's 2 "Unknown" nodes and it's impossible to use this worflow without them. AND THEY ARE NOWHERE IN THE WORKFLOW! So I can't bypass them. WTF?!