<h2 id="heading-1851">Cat memes its a Base</h2>I am glad to share with you a rofl motion lora which i trained just4fun, but he works unexpectedly good lolTrained on <a target="_blank" rel="ugc" href="https://youtu.be/xSE9Qk9wkig?si=gfy1aM-AgRZEQX5_">Keyboard cat meme</a> 16 frames 512x344mm_sd_v15_v2 , dreamshaper8 , 350 stepsKeywords: a cat play synthesizer5 hours on 3060ti 8gb<a target="_blank" rel="ugc" href="https://youtu.be/dQDJ1ZpKx4U?si=rqNKaqFpxaBl1jz_">Training guide</a><h2 id="heading-1852">The best settings I've found:</h2>a (username:1.1) play (DJ console:1.1),synthesizer ; and some vibe tokens like : chill zone,lofi vibe,ambient sound beats0.777 strength512x51264-80 frames8-9 steps with AnimateDiff Lightning 8 stepCFG 2.5Euler + sgm_uniformIp-Adapter-plus with weight_type "weak output" and "easy out"I understand that a weak output allows you to connect hands at the end of generation,because he doesn't look strong at references, but this is not a panaceaDon't forget to share your settings,it will help a lot to build better memes in future lol<h2 id="heading-1853">IpAdapter:</h2>This is very cool thing hat I always use with animatediff to set the styleBecause v2 understand the prompt style rly bad, it is much more profit to use a reference + short prompt + a little style promptThe task is to delicately give reference images with body&amp;hand pose , that will be easy to connectFor example, with hands up you have 99.99% chance to cause ghost hands because the model is trained with hands on the tableAdvise you to find references on Pinterest, also i created a <a target="_blank" rel="ugc" href="https://www.pinterest.com/aep_art/dj-mew/">board </a>with some DJ's, thanks to the community for sharing their pictures, I'll start doing it on pinterest too &lt;3You can combine and mix references whatever you want, feel free BroP.S. More experimenting with weight_types and Combine Embeds<h2 id="heading-1854">Resolutions:</h2>I made a mistake on training, and training was at 256x344 32 : 43 instead of 256x384 2:3, this is why u can achive some artifacts with default resolutions, uuupsWorks very good at 512x512, but I haven't experimented much more with resolutionsBut i also got some good results at: 512x384 , 512x640 , 640x512Note that if you want vertical or horizontal output with less artifacts, you need a similar aspect ratio references for IpAdapter<h2 id="heading-1855">Total frames imho:</h2>32 - maximum consistency, minimum motion (only one hand)48 - little more motion, but sometimes not enough frames to connect hands64&amp;80 - best results, with multiple poses and hand wigglingP.S. When you got a ghost hand, try add or remove 16 frames<h2 id="heading-1856">Roadmap:</h2>Train 2-3 flexing cats 384x256 (next step)Host Runpod and raise resolutionDo not use people for train (just4fun)Make more crazy motions lora<a target="_blank" rel="ugc" href="https://linktr.ee/aep.art">My Link Tree</a>

DJ Username motion lora

Trained on 16 frames <a rel="ugc" href="https://www.youtube.com/watch?v=xSE9Qk9wkig">keyboard cat</a>

Files with SHA256: c693522bf5c7be75bcbd9f4e079186e95bb55e60331ae70003a531a4c623f7bc