Hi~ I'm Juaner.This is a test model for training Flux fonts, with a dataset of 10 images and a total of 600 training steps. The main style is rounded, neon-colored fonts. On an RTX 4090, the total training time was 20 minutes. I will share the dataset on <a target="_blank" rel="ugc" href="https://huggingface.co/juaner0211">https://huggingface.co/juaner0211</a>.The training tool is ComfyUI, which is truly an extremely convenient training method!!! Thanks to hzwh88 for sharing. You can check out the LoRA training node here: <a target="_blank" rel="ugc" href="https://civarchive.com/models/690348">https://civarchive.com/models/690348</a>When testing the model, I found that Flux has an amazing ability to control styles. At 150 steps, it had already started to produce a style close to our dataset, and at 600 steps, the quality was already great. If you're interested in this area, feel free to discuss and explore with me. Let's discover more possibilities of Flux together!!!<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/3035616a-13fb-4600-a04d-418546301e65/width=525/3035616a-13fb-4600-a04d-418546301e65.jpeg" />Recommended prompt:Natural language + trigger wordFor example: Juaner_font, Theme: Night Vibes Text Content and Position: "Night" at the top, "Vibes" below, both in large glowing neon letters. Visual Content: A cool cartoon DJ spinning records, with music notes floating around. Background Description: A dark cityscape with bright neon signs and glowing streetlights.Recommended base model: flux devRecommended weight: 0.6-1Recommended size: 1024*1024Please adjust flexibly based on the results.Have fun, welcome everyone to share their generated images.If you have any commercial licensing or model customization needs, you can contact: <a target="_blank" rel="ugc" href="mailto:juaner0211@163.com">juaner0211@163.com</a>ღ( ´･ᴗ･` )Cheershi~我是卷儿。这是一个测试flux字体训练的模型，数据集共10张，训练总步数600步，主要风格是荧光色的圆圆的字体。在4090上总训练时长20分钟，数据集我会分享在<a target="_blank" rel="ugc" href="https://huggingface.co/juaner0211。">https://huggingface.co/juaner0211。</a>训练工具是comfyui，这个真的是超级方便的一种训练方式！！！感谢hzwh88的分享，lora训练节点看这里：<a target="_blank" rel="ugc" href="https://civarchive.com/models/690348">https://civarchive.com/models/690348</a>测试模型的时候发现flux对于风格会有非常惊人的把控，在150步的时候就已经开始出现我们数据集接近的风格，而600步的时候质感已经很棒。如果你有这方面的兴趣，欢迎来跟我一起讨论探索，我们一起发掘flux更多可能性！！！<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/f0d2d7a8-a2f6-4df5-9530-0aa547550302/width=525/f0d2d7a8-a2f6-4df5-9530-0aa547550302.jpeg" />推荐prompt：自然语言+触发词比如：Juaner_font, Theme: Night Vibes Text Content and Position: "Night" at the top, "Vibes" below, both in large glowing neon letters. Visual Content: A cool cartoon DJ spinning records, with music notes floating around. Background Description: A dark cityscape with bright neon signs and glowing streetlights.推荐底模：flux dev推荐权重：0.6-1推荐尺寸：1024*1024请根据结果灵活调整have fun, 欢迎大家返图。如果有商业授权或者模型定制需求可以联系： <a target="_blank" rel="ugc" href="mailto:juaner0211@163.com">juaner0211@163.com</a>ღ( ´･ᴗ･` )比心

Files with SHA256: e60cd24233b0ec0a5a89d6b0b377c16c9459e5da5ab10de58d1225d1b7392aea