<h3 id="v1.9">V1.9</h3>(Turbo included: CFG 1.2, Steps 16, I like ER_SDE/simple)Not yet the final form! This version is an upgrade of the Failed Experiment to get back on track with this model version. To do so, i leveraged <a target="_blank" rel="ugc" href="https://civarchive.com/models/2409949/sam-anima-realistic?modelVersionId=3074972">Sam ANIMA Realistic Turbo V2.3</a>. First i converted it to ANIMA 2.9B, then i did some level of merge into Failed Experiment:<pre><code>&gt;&gt;&gt; import sd_mecha as sdm
... a = sdm.model("UnNamedA29-FailedExperiment.safetensors")
... b = sdm.model("samANIMARealistic_turboV23-A29.safetensors")
... c = sdm.slerp(a,b,alpha=0.3)
... d = sdm.slerp(a,b,alpha=0.5)
... e = sdm.n_average(a,c,d)
... sdm.merge(e,output="alpha.safetensors",output_dtype="bfloat16")</code></pre>After that, it was some touch and go with the various upgrader LoRAs i had made previously for UnNamedA 2B, converted to 2.9B (including one based on muKrea).I also built a custom Qwen3-0.6B TE based on the base TE + my Qwen-Heretic LoRA (at 0.2) + DARE/TIES (at 0.3) from <a target="_blank" rel="ugc" href="https://civarchive.com/models/2673444/qwen-06b-uncensored-anima?modelVersionId=3001834">the uncensored TE</a> by @Felldude (i didn't use it as-is as it lost a bit of ANIMA knowledge from my test). This TE (together with a VAE) is baked in the checkpoint and will not be published as a standalone.V2 when i can set up a better training workflow!<a target="_blank" rel="ugc" href="https://civarchive.com/images/147785704">Version comparison: Failed vs V1.9 vs source SAM model</a><h2 id="failed-experiment">Failed Experiment</h2>12h (on 18) in my training of V1, i had a power outage... so i took a minute to check where i was. Epoch 9... of a failed training. Because, i started the training on the Turbo version of the model instead of the normal one. Let's say it messed up a few things 😅Not wanting to throw it away, i managed to salvage stuff and here we are. An INT8 Convrot standalone version of this model. It kinda works, but not in the way I wanted. Works like V1, is not AIO but with Turbo. And gives better results with Euler A / Simple.<h2 id="v1">V1</h2>This model is a finetuning of my previous conversion to ANIMA 2.9B of <a target="_blank" rel="ugc" href="https://civarchive.com/models/2697655/unnameda-anima-realistic-model">UnNamedA</a>.<h3 id="usage:">Usage:</h3><ul><li>/!\ Turbo model: CFG 1~1.5, Steps 8~16</li><li>ER SDE / Beta or euler A / sgm uniform</li><li>Positive prompt: masterpiece, best quality, highres, realistic, &lt;your prompt&gt;</li><li>Negative prompt: worst quality, low quality, blurry, anatomical nonsense, text, watermark, signature (optional: poorly drawn face, mutated face, ugly, cropped, out of frame, 3d, cgi, render)</li><li>Flow Shift: 3</li><li>HiRes.Fix: denoise 0.3, same CFG, same number of steps has standard pass, <a target="_blank" rel="ugc" href="https://civarchive.com/models/141491/4x-nmkd-superscale">4xNMKD Superscale</a></li><li>AIO: This model contains the VAE and TE</li></ul><h3 id="build:">Build:</h3>Unlike classic ANIMA models, <a target="_blank" rel="ugc" href="https://civarchive.com/models/2697655/unnameda-anima-realistic-model">UnNamedA</a> also has an heritage from <a target="_blank" rel="ugc" href="https://civarchive.com/models/2674046/cosmos-predict2-anima-architecture">Cosmos 2B</a> which brought some "glitches" when converted (some strange change in the face and a "grain" on the pictures). That's why i de-turboed UnNamedA 2.0, <a target="_blank" rel="ugc" href="https://civarchive.com/articles/36274/converting-a-2b-anima-model-to-a-29b-anima-model">converted it to 2.9B </a>and did a new pass (two actually) of finetuning directly on the 2.9B version using <a target="_blank" rel="ugc" href="//github.com/gazingstars123/Anima-Standalone-Trainer">this tool.</a>After some testing and merging to avoid outliers, i re-merged at 0.7 a new turbo (it really gives better results): the original cosmos2 turbo to avoid bringing anime aesthetic in the mix was converted into 2.9B via <a target="_blank" rel="ugc" href="https://github.com/storyAura/Anima2.9B-Lora-weight-conversion">this script</a> and then cleaned of incompatible keys.Here is a quick comparison:<ul><li>UnNamedA V2.0t-29B</li></ul><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/3bc312d1-5ae4-45fc-906c-eb9a7bb0a670/original=true/3bc312d1-5ae4-45fc-906c-eb9a7bb0a670.jpeg" /><ul><li>UnNamedA29 V1</li></ul><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/86d68fc2-ec3c-462a-b141-fe3ed1e40852/original=true/86d68fc2-ec3c-462a-b141-fe3ed1e40852.jpeg" />The "grain" can be seen all over the hair and skin on the previous version and the teeth where a bit strange.NB: Yes, I'll do a V2 to further train this model, since i feel it lost a bit of "texture" in the process (and this time, i'll do a proper teacher/student training).PS: A V1.5 is undergoing training:<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/993b5023-484a-4847-bb77-62c057214071/original=true/993b5023-484a-4847-bb77-62c057214071.jpeg" />GPU goes Brrrr (64 dim LoRA, 1536px resolution, i am nearing what i can train locally)<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/6510d884-27b8-4dd5-bb6b-df5761f78048/original=true/6510d884-27b8-4dd5-bb6b-df5761f78048.jpeg" />~8h done, ~10 to go!<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/4cdfba30-0f4d-48e5-a576-5581994aad8f/original=true/4cdfba30-0f4d-48e5-a576-5581994aad8f.jpeg" />Slowly going doooowwwnnnn...<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/e0e8f94e-b0e1-4ea0-b61f-675d1a94b76d/original=true/e0e8f94e-b0e1-4ea0-b61f-675d1a94b76d.jpeg" />Thanks for reading! 💜

UnNamedA29 - ANIMA 2.9B realistic model (Turbo/AIO)

Posted on Civitai, upload found elsewhere are FORBIDDEN and done without the author permission.

Files with SHA256: 477c626835acc7c8b3651ba3d93e3554bf803dcba93947a5b714a44b47bfae10