<h1 id="flux.1-dev-merged-with-5-loras.-clip-and-vae-include.-z8k87m2mp">Flux.1-dev merged with 5 loras. Clip and vae include.</h1><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/873b37f3-ecfb-4d23-b7c1-86adf4470dc6/width=525/873b37f3-ecfb-4d23-b7c1-86adf4470dc6.jpeg" />Workfow:<a target="_blank" rel="ugc" href="https://civarchive.com/models/673113/flux1-dev-merged-model-workflow-without-unet-loader?modelVersionId=753509">Flux.1-dev merged model workflow without Unet loader - v1.0 | Stable Diffusion Workflows | Civitai</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/688102/flux-with-ultimate-sd-upscale">Flux with Ultimate SD Upscale - v1.0 | Stable Diffusion Workflows | Civitai</a>Few words:I've been getting asked a few times: "Hey, your model generates perfect images, just like regular FLUX.1-dev, but it takes up more space... So why use it?"The answer isn't so straightforward. 99% of models based on the official DEV version create amazing images, but anyone who's ever created a checkpoint knows it's done for a specific reason.What's mine? In my case, it's the desire to avoid using LoRA to generate images in different themes. My favorite areas are science fiction, not necessarily cyberpunk or steampunk, hard/dark fantasy, and surrealism. So I created LoRA in these themes and combined them with 1-DEV. Additionally, my comfyui scripts are built so that I don't have to worry about loading CLIP_L, T55XXL - they're already included in the checkpoint. Plus, it's a FULL model, not trimmed like the smaller PRUNNED version. Why not trim it? I was waiting, maybe you were too, for the open-source world to catch up with MJ... That happened with the release of FLUX! I'd been saving up for almost a year to buy a card from the XX90 series. Since I already bought such a "powerful" piece of hardware, I don't want to limit myself and cut it down to the level of schnell, or fp4 or fp8. I want to use the full power of my GPU. Of course, I realize that my model is "a bit" larger than the standard one. That's why I uploaded it to tensor-art as a ready-to-use checkpoint, as well as an online application - for image generation and 2x upscaling. The second application serves as img2img for changing the background. If someone has the time and energy, they can easily check that my images (the same prompt, seed, sizes) generate completely different images than pure 1-DEV.At this point, I want to express my immense gratitude to all the donors (who contributed funds for early access, and sent me support directly to my Civitai and TensorArt accounts). Your support was a huge motivation for me to create an even better version - v2. Thank you! Have fun with my models.<a target="_blank" rel="ugc" href="https://tensor.art/u/689683550856812161/templates">Aderek514's AI Tools | </a><a target="_blank" rel="ugc" href="http://Tensor.Art">Tensor.Art</a><a target="_blank" rel="ugc" href="https://tensor.art/u/689683550856812161/models">Aderek514's Models | </a><a target="_blank" rel="ugc" href="http://Tensor.Art">Tensor.Art</a>Workflow to v5 fp8 version : <a target="_blank" rel="ugc" href="https://openart.ai/workflows/macaque_keen_26/zz0jLfIYb0fdITK4ZHpQ">Aderek Flux with XLab sampler | ComfyUI Workflow (</a><a target="_blank" rel="ugc" href="http://openart.ai">openart.ai</a><a target="_blank" rel="ugc" href="https://openart.ai/workflows/macaque_keen_26/zz0jLfIYb0fdITK4ZHpQ">)</a>

Aderek Flux

In this version i remove CLIP and VAE.

aderekFlux_withoutClipVae.safetensors

without clip&vae

V6 fp8 Halloween Edition

v5 fp16

v5 fp8

fp8 v4

fp8 v2

Schnell fp8

v2.0

v1.0

Files with SHA256: d22448fee2f5ff5c2e53d194c9a91b074302f2174c70dd756c043b3453376e82