Hi! I'm a Video Plus API subscriber I'd like to request a custom model installation for WAN 2.2 image-to-video: CHECKPOINT (main request): DaSiWa WAN 2.2 I2V 14B Lightspeed (fp8 safetensors) https://civitai.com/models/1981116 Version: "SnatchKiss High v11" (plus its matching Low-noise file if applicable) → Ideally exposed as an i2v workflow (like animate-smoothwan22) so it can be used with LoRA concepts and the /size: parameter. LORAS (all WAN 2.2 I2V, high+low noise pairs where available): Bouncing Boobs: https://civitai.com/models/1343431 Deepthroat/Blowjob: https://civitai.com/models/1497390 DR34ML4Y All-In-One v2: https://civitai.com/models/1811313 Slop Bounce: https://civitai.com/models/1944129 Slop Twerk: https://civitai.com/models/2008663 POV Paizuri/Titfuck: https://civitai.com/models/2070335 Teasing Sensual Blowjob: https://civitai.com/models/2231076 Thanks a lot! Also: is there a parameter to control output FPS/frame count on animate workflows, or a recommended way to run frame interpolation (16→32fps) on your side?

HSIU YUAN CHEN 7 days ago

1

I cannot code, and therefore I cannot even begin to suggest how to do this, but i do use the Krita AI plug in, which has already implemented a similar feature. It’s open source code. If the webui is running ComfyUI in the background, then perhaps the code will be similar? I suppose this will also require insightface to be installed too. Here is the link for the plugin that has this feature. https://github.com/Acly/krita-ai-diffusion

Jason Lee almost 2 years ago

A way (at least on webui/stable fo go) to import using download link. On another service I use you can import from a few places and they are added to your model lists

PanofArcadia almost 2 years ago

2

SDXL isn't that great with 512x512 and there isn’t that big performance hit if the default for these were more suitable 768×768 or even 1024×1024. And SDXL models are fully capable handling those resolutions without hallucinating duplicates and other resolution based glitches.

Hampsteri almost 2 years ago

/wf command as alias of /workflow /redo, /re command for doing again job at the quoted message. That command must to delete /seed variable to not re make same images. (Sorry about my English level)

İrfan Toğluk almost 2 years ago

Disclaimer is I'm dumb at telegram, but a magnific style upscaler for would be killer. Prompt or no, weight or no. Pic style, any image, as refined as your machine can handle

ameer whitmyer about 2 years ago

1

A studio style animator in gradient would be so good. Of course it turns into frame to hang more tools on...convert files, spot upscale, brush editors, morph body/head/ arms/legs...big dream, but there are a lot out there sygmoid is open source im sure others, if we had a pirate free express interface a lot of us would be happy to leave adobe and others I thinkterface

ameer whitmyer about 2 years ago

1

To have an interactive prompt builder mode, eg: select "/brew", "/render", etc to have an easier time building a prompt before rendering. Pros: Easier onboarding for new users and easier introduction of new features or modes for experienced users Cons: Telegram may be buggy. May be redundant if PD pushes to an external app.

Scorpinaus about 2 years ago

/more and /facelift works good on Telegram, but we doesn't have a good option on web versions (Stable2Go, Unified and etc...). I think if we have a way to set weights on Adetailer like "/adetailer /ad_stregth:0.4" would be awesome. For Hires.Fix could be variables to Scale (multiplier, limiting from 1.1x to 2.0x), Steps and Stregth, like: "/hires:2.0 /hires_steps:13 /hires_strength:0.7". I believe that by adding this the software would simply be perfect. There would be nothing more to be done in my opinion <3</p>

Raphael Reis about 2 years ago

not only in the modal window

Tomato about 2 years ago

For example if i want to see all base models, but no pony-models?

Tomato about 2 years ago

This would be a significant lift, as we would have to replicate all of the social functions of Telegram, but it would also give us more control and stability not bound by Telegram rules or outages. Of course, it would have to be side-loaded as the PD experience would not fly on the Google Play store at all, meaning it may be difficult to get updates (or) it could be done as a web app, but it wouldn't be as punchy and fast as a real app. Discuss and upvote if you want to see a PD app, I think those are the pros and cons.

Captain about 2 years ago

RonRider noticed that if there are two spaces ahead of the concept, or spaces between bracket model names, the facelift and highres commands don't run

Captain about 2 years ago

The feature is very fast but most people don't know that it works even better with /size. As we have speed efficiency already, let's up the resolution by default

Captain over 2 years ago

1

/meme top:I feel it ;;; /meme bottom:I feel the cosmos

Captain over 2 years ago