<h2 id="moss-or-jima-style-lora">Moss | Jima Style Lora</h2>Base Model:<a target="_blank" rel="ugc" href="https://huggingface.co/OpenMOSS-Team/MOSS-Image-Base">Moss-Image-Base-bf16</a>I've uploaded three models with different levels of training. Personally, I recommend using the version with 3,500 steps, but you can adjust it yourself if you want to experiment. Content containing NSFW imagery may be subject to some censorship; you can use prompts to minimize this effect. Generation settings<ul><li>Works at resolutions between 512^2 and 1536^2 pixels.</li><li>No <a target="_blank" rel="ugc" href="https://huggingface.co/lodestones/MOSS-Image-Base-8-steps-dmd">dmd</a> 30-50 steps, CFG 3-5</li><li>+ <a target="_blank" rel="ugc" href="https://huggingface.co/lodestones/MOSS-Image-Base-8-steps-dmd">dmd</a> 8 steps, CFG 1</li><li>A variety of samplers work. Some of my favorites:</li><li>Euler + SGM Euler + Simple</li></ul>I recommend using Base + dmd2 distillation acceleration for inference; it tends to perform better than SFT, but this is just a suggestion—it depends on your personal preference.Quoted from the official description; you may also refer to the parameter settings shown in the example image<h1 id="krea-2-or-jima-style-lora">Krea 2 | Jima Style Lora</h1>BASE Model:Krea 2 RawTraining on ModelsSope<h1 id="anima-orjima-style-lora"> Anima |Jima Style Lora</h1>base Model:<a target="_blank" rel="ugc" href="https://civarchive.com/models/2458426/anima?modelVersionId=2945208">Anima Base-v1</a> Generation settings<ul><li>Works at resolutions between 512^2 and 1536^2 pixels.</li><li>30-50 steps, CFG 4-6.</li><li>A variety of samplers work. Some of my favorites:<ul><li>er_sde: neutral style, flat colors, sharp lines. I use this as a reasonable default.</li><li>euler_a: Softer, thinner lines. Can sometimes tend towards a 2.5D look. CFG can be pushed a bit higher than other samplers without burning the image.</li><li>dpmpp_2m_sde_gpu: similar in style to er_sde but can produce more variety and be more "creative". Depending on the prompt it can get too wild sometimes.</li></ul></li><li>If going for a more realistic / painterly look, the beta57 scheduler (ComfyUI RES4LYF custom node pack) can help make better textures, since it puts more emphasis on low-noise timesteps.</li></ul>Quoted from the official description; you may also refer to the parameter settings shown in the example image

Jima Style Anima&Krea 2&Moss

jima-comfy-3500.safetensors

Files with SHA256: 0c1ec9063f7eddd0b34ebeb8637882925ffe749d9497825f4a4b765123ac33e3