Sponsorship: For sponsorship inquiries, please reach out to us on Discord. Your support helps us grow and improve our future work.
Buy Custom Models: Interested in purchasing a custom LoRA model? Contact us we can deliver quickly.
Epic Vision
I'm thrilled to introduce Epic Vision, a model crafted to create stunning, fantasy-inspired images. This model is designed for those looking to add a bit of magic and grandeur to their visual creations. Whether you're creating epic landscapes, fantastical characters, or anything in between, this model will add a touch of fantasy style to every output.
Model Details:
Type: Fantasy-style image generation
Trigger Words: None required, but using fantasy-related prompts can enhance results
Output: High-detail, imaginative, and fantastical images with a unique blend of artistic elements
NSFW Support: No
How I Made This Model:
I trained some LoRA models and then merged them with the Flux dev fp8 model. During this process, I made a few optimizations. The model now has CLIP and VAE baked in, so you don’t need to use separate versions.
Quick Parameters:
Sampler: Euler
Scheduler: Simple
Sampling Steps: 20
CFG Scale: 1
Join my Discord community to share your creations, provide feedback, and explore the world of Epic Vision with fellow enthusiasts! 😊
Description
FAQ
Comments (4)
We REALLY need to get into the habits of not adding the T5 and VAE to every new model. There's no need. Its just extra work for you and less HD space for us.
Edit: Super nice model though :) 10/10 IMO
yes please lets stop making this the norm. at least add the base model alongside the AIO.
@colinw2292823 Thanks for the feedback! 😊 I actually uploaded two versions of the model to give users the choice. The version you're commenting on includes the baked CLIP and VAE, which makes it around 15GB. But there's also a lighter version without the CLIP and VAE that’s only 11GB, for those who already have them installed or prefer the smaller size. Hope that helps! 🙌
@colinw2292823 @soupman1 @Ai_Art_Vision
Colleagues there is a small but very annoying issue with using Dual Clip Loader node in ComfyUI.
If you use a GGUF loader or Diffusion Model Loader with Dual Clip Loader on RTX 3060 w/ 12GB of VRAM then the first run with changed prompt text you get an Out Of Memory error (I'm getting at least).
For example, I can't use Queue (instant) feature to add a new prompt whenever queue gets empty. ComfyUI just stops on OoM error.
But when I use a regular Checkpoint Loader from comfy.Core and use only the CLIP module, then CLIPs memory just getting correctly handled and changed Prompt Text no longer generates the Error.
So, I personally WANT to have a choice of a ckpt WITH CLIP parts integrated.
VAE is not really needed though.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.