Sun]flower
<p>Hi!</p><p>Sorry for the lack of news but I have been doing lots and lots of training so I can understand better SDXL new architecture. I'm on it but for now I wanted to present (and after 2 weeks of heavy tunning) <strong>Sun]flower</strong>, a new SDXL LORA and my very first SDXL Style LORA.</p><p><br />Sun]flower is a SDXL LORA that has <u>certain sunflower plant aesthetics</u> and can be used in a lot of different ways. I have tested it in WebUI and ComfyUI... I'm not sure yet which one I liked most but the sample images I put here (except 2) are from ComfyUI, in the next days I will upload some Webui Sample images too.</p><p><br /><br />- If you want to know <strong><u>how to prompt</u> please <u>refer to the sample images</u>.</strong></p><p>- <strong>Trigger</strong> is <strong>snfw3r</strong></p><p><strong>- </strong><u>Recommended weight is </u>1 but if you feel <u>there are too much sunflower in the scene use 0.6 onwards</u>. Depends a lot on the concept and seed.<br />- <u>Resolution recommended 1024x1024 or 1024x768/1024x1456</u> and even 512x768 or 768x768. Resolution will guide too the prompt, remember this.</p><p>- You can guide more the AI towards the style using "3d render", "drawing", etc. Remember that now SDXL will follow much better prompt.</p><p>- This LORA is <u>good for almost anything except probably realistic</u> but try to maintain prompts simple as possible and focus on what you want to see.</p><p>- This LORA was <u>trained in more than 1 style so it can transfer</u> to concept, illustration, 3D render, etc. Guide it with your prompt.</p><p>- This LORA <u>wasn't testted on finetunned XL</u> Dreambooth models because I wanted to see the inner strenght of it. Feel free to test it wherever you want.</p><p></p><p>In the case you want my workflow:<br /><a target="_blank" rel="ugc" href="https://huggingface.co/datasets/davizca87/C01nWorkflow/tree/main">https://huggingface.co/datasets/davizca87/C01nWorkflow/tree/main</a> <strong><- LOAD it via Json file in Comfyui</strong></p><p><br /><br />And that's all. As always, eternal gratitude for all people giving feedback here, in Discord and of course supporting us in <a target="_blank" rel="ugc" href="https://www.patreon.com/user?u=90661954">Patreon</a>!</p><p>Remember we are in SDXL territory and I'm not as versed as in SD 1.5 so feeback, positive or negative is very much welcome.<br /></p><p>NOTE: I'm currently studying Prodigy IA3 LOKR but for now you will habe to deal with this adafactor optimizers.</p>
Image examples for the model:
