I was having a hard time consistently creating images with characters wearing a white fitted crop top. Therefore, I've decided to train one Lora model for it.<h2 id="currently-the-lora-works-best-using-the-following-settings:">Currently, the Lora works best using the following settings:</h2>Trigger word: chrnwfct Additional / useful prompts (positive): white fitted crop top, short sleeves, crew neck, plain, solid color, unbranded, blank shirt (*if you're going for the same plain white look) Additional / useful prompts (negative): long sleeves, hoodie, jacket, coat, cardigan, dress, tank top, camisole, spaghetti straps, pattern, print, (logo:1.4), (text:1.4), (graphic:1.3), (print:1.3), (emblem:1.2), (brand:1.2), (pattern:1.2), stripes, lace, mesh, sheer, turtleneck, collar, v-neck, off-shoulder, bow, baggy, loose fit, ruffles, solid color, unbranded, blank shirt (*adjust as you see fit) Weightage: 0.8 ~ 1 (it's best to use 1, though I've tried 0.8 and the output isn't so bad. I don't like the output at 0.6) Checkpoint: <a target="_blank" rel="ugc" href="https://civarchive.com/models/967405/nova-orange-xl">Nova Orange XL</a> (it's the only one I've used so far to test the outputs) <h2 id="weaknesses">Weaknesses</h2>The Lora was trained using a small set of images as I couldn't find other images of the same kind of white crop top I was looking for (← I'm a little picky about this). Additionally, I've only tested using the <a rel="ugc" href="https://civarchive.com/models/967405/nova-orange-xl">Nova Orange XL</a> checkpoint and not others, since it takes quite some time to generate images. My PC is not that strong...<h2 id="conclusion">Conclusion</h2>So far, I've generate about a dozen of images after tweaking the prompts and outputs have been consistent. I'm planning to train more Loras for other types of clothing once I've reached the required Buzz amount.I look forward to your feedback on the Lora, and if possible, if I find more images of the same kind of outfit, or the time to generate more outputs of the same kind of outfit, I'll retrain the Lora.Have fun generating! 😊

White Fitted Crop Top

2115402_training_data.zip

White_Fitted_Crop_Top.safetensors

WFCT (Illustrious) v1.0

THIS IS A <a href="https://civitai.com/models/1868931" alt="Reprint url">REPRINT</a>!Go to original author to check their other models or to support them. <hr /> I was having a hard time consistently creating images with characters wearing a white fitted crop top. Therefore, I've decided to train one Lora model for it.<h2 id="currently-the-lora-works-best-using-the-following-settings:">Currently, the Lora works best using the following settings:</h2>Trigger word: chrnwfct Additional / useful prompts (positive): white fitted crop top, short sleeves, crew neck, plain, solid color, unbranded, blank shirt (*if you're going for the same plain white look) Additional / useful prompts (negative): long sleeves, hoodie, jacket, coat, cardigan, dress, **** top, camisole, spaghetti straps, pattern, print, (logo:1.4), (text:1.4), (graphic:1.3), (print:1.3), (emblem:1.2), (brand:1.2), (pattern:1.2), stripes, lace, mesh, sheer, turtleneck, collar, v-neck, off-shoulder, bow, baggy, loose fit, ruffles, solid color, unbranded, blank shirt (*adjust as you see fit) Weightage: 0.8 ~ 1 (it's best to use 1, though I've tried 0.8 and the output isn't so bad. I don't like the output at 0.6) Checkpoint: <a target="_blank" rel="ugc" href="https://civitai.com/models/967405/nova-orange-xl">Nova Orange XL</a> (it's the only one I've used so far to test the outputs) <h2 id="weaknesses">Weaknesses</h2>The Lora was trained using a small set of images as I couldn't find other images of the same kind of white crop top I was looking for (← I'm a little picky about this). Additionally, I've only tested using the <a rel="ugc" href="https://civitai.com/models/967405/nova-orange-xl">Nova Orange XL</a> checkpoint and not others, since it takes quite some time to generate images. My PC is not that strong...<h2 id="conclusion">Conclusion</h2>So far, I've generate about a dozen of images after tweaking the prompts and outputs have been consistent. I'm planning to train more Loras for other types of clothing once I've reached the required Buzz amount.I look forward to your feedback on the Lora, and if possible, if I find more images of the same kind of outfit, or the time to generate more outputs of the same kind of outfit, I'll retrain the Lora.Have fun generating! 😊

Files with SHA256: 1f5b02d1378b06dac42a5bd1124434e1812ac60913125f15b1c8dd9dc9e6c434