TL;DR: This is a DMD2 LoRA = use CFG 1 and 8-12 steps (even if i personally prefer 16 steps, that's still half of what i usually use and i get my pictures in 30s instead of 80s).Quick clarification:<ul><li>V1 is using turbo 0.1</li><li>V2 is using turbo 0.2</li><li>V1.5 is using a merge of 0.1 and 0.2</li><li>V3 is based on anima-turbo-v1.0</li><li>V2.9 is based on anima-turbo-v1.0 and turbo 0.2</li><li>All of them has been merged with the revised cosmos2.5 DMD2 LoRA</li><li>Yes, you can use CFG 1.2, 1.4, etc... you may need to add more steps then (that's why i do 16).</li></ul><h3 id="v2.9">V2.9</h3>With more extensive tests and help from heavy user of the previous version, it was clear that V3 was killing LoRA when used. So, here is V2.9, a step back to previous turbo for the sake of a better compatibility with LoRA. Recipe (merge done at Rank 128/64):<ul><li>Cosmos2.5 DMD2 LoRA: 0.3</li><li>Extract Turbo V1.0 LoRA: 0.2</li><li>Turbo v0.2 LoRA: 0.4</li></ul>Then, all this went through my <a target="_blank" rel="ugc" href="https://civarchive.com/articles/32545/lora-compression-dynamic-rerank-per-layer">compressor </a>(MAXDIM 64, CUTOFF 98) to get this slightly heavier LoRA, but providing from my first tests better results. 🥰<h3 id="v3">V3</h3>With <a target="_blank" rel="ugc" href="https://civarchive.com/models/2458426/anima?modelVersionId=3108589">Anima-turbo-v1.0</a> officially out, i knew some people would ask about an update. So here it is! I switched my approach here as this is a full model we are talking about. First, i merged base and turbo at 50/50 to get "half-turbo" in which i merged the Cosmos2 LoRA at 0.5 to get "custom-turbo". V3 is a LoRA extraction of "custom-turbo" minus "base".<h3 id="v1.5">V1.5</h3>Borrowing a book from @kai and for the people not satisfied by 0.1 and 0.2 and wanted something in the middle, i did a merge of official 0.1 and 0.2 before doing the merge with the cosmos2.5 distilled LoRA. This end being a version between V1 and V2 = V1.5. This will probably be the last version except if a 0.3 version of the official LoRA comes out (and/or if a full distilled version of ANIMA base is released). NB: this version feels more sensitive to watermarks, you can use CFG 1.2-1.4 to be able to add a negative.<h3 id="v2">V2</h3>By popular demand, here is a V2 of this LoRA based on the V0.2 of the source turbo LoRA. 😘<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/b07e33f2-c915-4864-ac41-185fdbaf911d/original=true/b07e33f2-c915-4864-ac41-185fdbaf911d.jpeg" /><h3 id="v1">V1</h3>This is my "personal" brand of the classic ANIMA Turbo LoRA, based on the <a target="_blank" rel="ugc" href="https://arxiv.org/abs/2405.14867">DMD2 </a>method.There is already several "Turbo" LORA for ANIMA:<ul><li><a target="_blank" rel="ugc" href="https://civarchive.com/models/2560840/anima-turbo-lora">Anima Turbo LoRA</a></li><li><a target="_blank" rel="ugc" href="https://civarchive.com/models/2364703/rdbt-anima">RDBT Anima</a></li></ul>But interesting enough, <a target="_blank" rel="ugc" href="https://civarchive.com/user/polyx">someone </a>had the idea to make and test a LoRA using the original Cosmos-Predict2.5 architecture: <a target="_blank" rel="ugc" href="https://civarchive.com/models/2466415/cosmos-predict25-2b-base-distilled-extracted-dmd2-lora">https://civarchive.com/models/2466415/cosmos-predict25-2b-base-distilled-extracted-dmd2-lora</a>The main process was: extract (using SVD) a LoRA out of "<a target="_blank" rel="ugc" href="https://huggingface.co/nvidia/Cosmos-Predict2.5-2B/tree/main/base/distilled">Distilled Cosmos</a>" minus "<a target="_blank" rel="ugc" href="https://huggingface.co/nvidia/Cosmos-Predict2.5-2B/tree/main/base/pre-trained">Original Cosmos</a>" and test it on ANIMA... and it works! (even if some keys are refused because they changed shape in ANIMA). Below is a comparaison of what those bring to the table (RDBT is very close to the base Turbo and so was excluded).<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/82ed98cb-f7b3-440a-8f5c-d6c53c4b3d02/original=true/82ed98cb-f7b3-440a-8f5c-d6c53c4b3d02.jpeg" />Now, what is did is "stupid" = Merge those two 😁 (i left RDBT alone as sharing merge is forbidden).To do so, i re-keyed both LoRA and do a re-rank of the resulting merge (32/16 instead of 64/64+32/32). The ratio was 6/4.This way, i get more details but with a less saturated result.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/f300a3ca-f89c-41c2-9b2f-e130f48aff39/original=true/f300a3ca-f89c-41c2-9b2f-e130f48aff39.jpeg" />And it works perfectly with <a target="_blank" rel="ugc" href="https://civarchive.com/models/2600825/narnima-personal-style-for-anima?modelVersionId=2937058">nArnima V2</a>:<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/3659d8ae-d338-42dc-8818-2fd1893f755d/original=true/3659d8ae-d338-42dc-8818-2fd1893f755d.jpeg" />Thanks for reading! 💖<hr />Legal<span style="background-color:rgb(26, 27, 30);color:rgb(141, 141, 143);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">The CircleStone Model is licensed by CircleStone Labs LLC under the CircleStone Non-Commercial License. Copyright CircleStone Labs LLC. IN NO EVENT SHALL CIRCLESTONE LABS LLC BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.

TURBO for ANIMA (less steps)

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

Files with SHA256: a5135d1cb868d4b3914cca6b32b1354e748c90488de1ab44a7dd7ea6ad003475