Workflow Venice AI Workflows and Models (DON'T POST REAL PEOPLE!!)

I'm getting Failed to fetch (outerface.venice.ai) while trying to use wan 3.0 ref, is it working for u guys ?
 
Seedance 2.5 costs about the same as WAN 3.0 even with the 20% discount. You just need to input a 1:1 sized starting image. You can crop your existing images to 1:1 to try it.

Seedance 2.5 (30s/480p)
217 credits (with 1:1 starting image)
289 credits (without 1:1 starting image)

WAN 3.0 (30s/480p) costs 213 credits. (20% discount)

But I am pretty sure WAN 3.0 will be on perma discount. Remember WAN 2.7 was also permanently discounted on Venice.

----------------------------------------------------------------
TBH, WAN 3.0 isn't really impressing compared to Seedance 2.5 so i don't really use it. Seedance 2.5 was a significant update compared to 2.0 regarding the NSFW performance, so I kept tinkering with it.

Waiting for ByteDance's new world model, which is expected to release sometime in October. Hopefully we will be able to interact with our goon videos.

https://x.com/business/status/2096940316431405420
 
WAN3.0 30sec@480p is currently 156 credits (195 at full price).

Even at full price you’re talking a $0.22 difference per-gen between SD2.5 and WAN3.0.

Don’t get me wrong, I like SD2.5 and have made some incredible stuff with it, but pretending like it’s not a money pit is a bit disingenuous.
 
I've been enjoying Wan 3.0 R2V simply for the fact that if you upload a vid where the subject is already talking, and it does a decent job at matching their voice. You don't have to upload a separate audio file.

This combined with the "reshape story" preset is pretty good in my opinion. So I prompt that the video be the same for the first 3 seconds, then have them randomly do or say something else in their voice.
 
Is there a venice competitor that offers the same models but that allows you buy and spend credits without forcing you to sub to generate nsfw?
 
Not everything on Wavespeed is NSFW friedly, at least on the website which has the mature filter set by default. However if you use the API, which is very easy to set, you can use it in comfyUI. I've switched over from Venice to Wavespeed, as it's cheaper.
 
Hi everyone,
I'm working on a photorealistic AI video project and need strong visual continuity between shots.
On Venice AI, the models that support image references seem to have stronger content restrictions, while some uncensored models don't support image references.
Is there any workaround to use an existing image as a reference with these uncensored models?
Any workflow or model recommendation that solves image reference + fewer restrictions would be really helpful.
Thanks!
 
Is anyone else not able to generate videos? I keep getting a timed out error message....could it be because I'm uploading NSFW content without signing up?
 
What do you guys use to source positions and stuff? I've been using old grok imagine gens but it can get limited. Do you just use whatever you come across in porn scenes?
 
Wan 3.0 reference to video results looks weird. Like everything else looks fine, but the face looks somewhat off. Like, it's the exact face sure, but it doesn't really do facial expressions well. It just looks copy and pasted whereas everything else is perfectly fine.
 
You must log in or register to reply here.
Back
Top