v.4 fixed the nodes that had been reset.
v.3 fixed the vae decode tile size. Thanks for pointing that out.
v.2 Streamlines things a bit. It should be easier and faster to use now.
The workflow supports loras, both double_blocks and clip support. Users can add a negative prompt. It uses TeaCache to speed up multipe generative upscale passes. Then it does a regular upscale and crops the image for presentation.
You can drag and drop the diagram or the gif to use the workflow without having to download the zip.
I wouldn't say that it's better than any other workflow in particular but people kept asking for it.
Description
streamlined
FAQ
Comments (11)
hi thanks for sharing.
I already saw that double conditioning attached to combiner for negative prompt in another workflow, wich i tested time ago and i coudnt find any reasons why that is built like that. i removed the combiner and one conditioning and attached the negative regulary and it workes the same way (as it should) am i missing something? any particular reasons why some users are doing that? i can't understand
hello, how do Install the module called "ApplyFBCacheOnModel", I can't find it anywhere
thanks. VAE tile size is 1536 on the first pass. is this a mistake ?
I haven't tried it out yet but I was wondering, the last block is called "upscale and interpolation" but I don't see an interpolation step, or am I missing something?
Hi man, is there any chance of having an img2vid workflow, congratulations on your work.
hey great job ! but i can't find this Lora.. de3pthr0at_v1.2.safetensors :(
Nice! But I'm confused how to use the upscale on the low res generation without starting a new generation in the low res pass. Would you please help me out?
Hi, thanks for this. What is the hjepoch99 lora? Is that available somewhere?
I can't seem to go longer than 61 frames without it going OOM after the first step. Any tips?
Can someone please explain the connections for LoRas in this workflow? I don't understand why they're connected they way they are. Also, can't seem to find this hjepoch99.safetensors :(
Why are some connected to Set_Model and some are not?
BTW thanks for the nice workflow, I could get stuff to generate, but I just don't understand LoRa chaining :D
The TeaCacheHuanYuanVideoSampler Nodes are still missing after installing the missing nodes.
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.