<h1 id="bento-is-a-carefully-designed-anime-flux-model-that-helps-produce-vibrant-dynamic-illustrations-in-a-variety-of-anime-inspired-styles-from-the-1980s-to-the-2020s.-bmkwfhffl">BENTO🍱 is a carefully designed Anime Flux model that helps produce vibrant, dynamic illustrations in a variety of anime-inspired styles from the 1980s to the 2020s.</h1>some things to know about your lunch:<ul><li>You do not need to specify whether an image is "anime" or "in the style of (anime studio)." Instead, describe the image and any effects you want to see and let the model deal with the styling.</li><li>artist/studio names do nothing (and this is on purpose) don't waste your time - if copying art styles is your deal, just add a Lora.</li><li>NSFW wasn't added beyond what flux dev had on release, SOME poses/concepts come through simply because of some fanservice stuff in the training data, but it's mostly SFW and should be used in that way. Obviously with some loras and a dream anything is possible still, but I want the default experience to be a common middle ground for everyone.</li></ul> Struggling with flux prompts? <a target="_blank" rel="ugc" href="https://chatgpt.com/g/g-oODh6sLdt-flux-prompt-crafter-by-bobsblazed">try my flux prompt craft GPT bot</a>Do you use the GGUF? Try these workflows. <a target="_blank" rel="ugc" href="https://civarchive.com/models/594149?modelVersionId=842917">GGUF T5 CLIP only</a> | <a target="_blank" rel="ugc" href="https://civarchive.com/models/594149?modelVersionId=842881">GGUF UNET+GGUF T5</a>Looking for a Latent node that's optimized for FLUX, SD3 &amp; SDXL? <a target="_blank" rel="ugc" href="https://github.com/BobsBlazed/Bobs_FLUX_SDXL_Latent_Optimizer?tab=readme-ov-file">Bob's FLUX Latent Optimizer Node</a>For optimal results, I recommend using 20-32 steps.Remember to also download the custom CLIP L! <a target="_blank" rel="ugc" href="https://civarchive.com/models/618792?modelVersionId=876968">The current version of CLIP L is V5</a>Want an alternative CLIP L? <a target="_blank" rel="ugc" href="https://huggingface.co/zer0int/LongCLIP-SAE-ViT-L-14">Try this one by Zer0int</a>

BENTO

Recommendations for [SERVING 4]:<ul><li>flux guidance: 2.7, 3.0, 3.5, 4.0</li><li>steps: 20-32</li><li>sampler: euler</li><li>guider: sgm_uniform</li><li>T5XXL: <a target="_blank" rel="ugc" href="https://huggingface.co/easygoing0114/flan-t5-xxl-fused/blob/main/flan_t5_xxl_full_FP8e4m3.safetensors">Flan T5 Merge</a></li><li>CLIP L: <a target="_blank" rel="ugc" href="https://huggingface.co/zer0int/LongCLIP-GmP-ViT-L-14/blob/main/Long-ViT-L-14-BEST-GmP-smooth-ft.safetensors">LongCLIP-GmP-ViT-L-14</a></li></ul>

bento_servingFOUR.safetensors

BENTO🍱

Serving:FOUR

Serving:THREE-GGUF

Serving:THREE

Serving:TWO-GGUF

Serving:TWO

Serving:ONE-GGUF

Serving:ONE

BENTO🍱 is a carefully designed Anime Flux model that helps produce vibrant, dynamic illustrations in a variety of anime-inspired styles from the 1980s to the 2020s.some things to know about your lunch:<ul><li>You do not need to specify whether an image is "anime" or "in the style of (anime studio)." Instead, describe the image and any effects you want to see and let the model deal with the styling.</li><li>artist/studio names do nothing (and this is on purpose) don't waste your time - if copying art styles is your deal, just add a Lora.</li><li>NSFW wasn't added beyond what flux dev had on release, SOME poses/concepts come through simply because of some fanservice stuff in the training data, but it's mostly SFW and should be used in that way. Obviously with some loras and a dream anything is possible still, but I want the default experience to be a common middle ground for everyone.</li></ul>Struggling with flux prompts? <a href="https://chatgpt.com/g/g-oODh6sLdt-flux-prompt-crafter-by-bobsblazed" rel="ugc noopener noreferrer nofollow" target="_blank">try my flux prompt craft GPT bot</a>Do you use the GGUF? Try these workflows. <a href="https://civitai.com/models/594149?modelVersionId=842917" rel="ugc noopener noreferrer nofollow" target="_blank">GGUF T5 CLIP only</a> | <a href="https://civitai.com/models/594149?modelVersionId=842881" rel="ugc noopener noreferrer nofollow" target="_blank">GGUF UNET+GGUF T5</a>Looking for a Latent node that's optimized for FLUX, SD3 &amp; SDXL? <a href="https://github.com/BobsBlazed/Bobs_FLUX_SDXL_Latent_Optimizer?tab=readme-ov-file" rel="ugc noopener noreferrer nofollow" target="_blank">Bob's FLUX Latent Optimizer Node</a>For optimal results, I recommend using 20-32 steps.Remember to also download the custom CLIP L! <a href="https://civitai.com/models/618792?modelVersionId=876968" rel="ugc noopener noreferrer nofollow" target="_blank">The current version of CLIP L is V5</a>Want an alternative CLIP L? <a href="https://huggingface.co/zer0int/LongCLIP-SAE-ViT-L-14" rel="ugc noopener noreferrer nofollow" target="_blank">Try this one by Zer0int</a>

Files with SHA256: 5c7f3068758c3d9d9c1605188266e23b368d1d30b264e6df07751d0c4be9ce6f