Izuku Midoriya MHA

Izuku Midoriya (Deku) | LoRA — My Hero Academia | Base: ANIMA

🙏 Before anything else: this LoRA pack took a lot of work — picking the images, building the dataset and tags, training, and testing the checkpoints. If you have a moment, drop a Like and especially post your results in the comments (with the prompt/settings you used) — it helps me a lot to know what to improve in the next versions. 🔥

⚠️ The model only uses images of Midoriya — that's why Bakugo didn't turn out well. Use it with respect.

💬 Check out how the prompt examples were set up in the posted images. 💬 Leave your generations in the comments, and tag the LoRA so I can see how it's looking, it would help a lot.

📦 A Midoriya (Izuku) LoRA pack trained exclusively on clean images. Releasing 5 checkpoints across two versions so you can pick the one that fits your project.

📊 About this model

This pack has two separate versions, each trained on a different image source — pick the one that fits the look you're going for.

🎨 Digital Art version — 175 images

Trained on digital art illustrations. Produces cleaner lines and tends to generate a younger, pre-adult Midoriya with some muscle definition. Strongly emphasizes the hero costume — specify a different outfit in the prompt if needed, otherwise the model defaults to the base hero suit.

📺 Anime version — 2400 images

No NSFW content was used in the training. Trained only with anime screenshots. Generates Midoriya in a more authentic anime style. More steps = stronger anime look, but the features also get heavier and noisier. All checkpoints produce a recognizable Midoriya. Looks younger and more cartoonish since only anime frames were used — great if you want a comic style reminiscent of the early episodes of the anime, with slightly off proportions and a nostalgic feel.

ℹ️ Generation tips

• On ANIMA base with 12 steps + acceleration LoRA, the output keeps a clean anime feel that works well for Midoriya — recommended as a starting point

• Using more steps or a different ANIMA-based model with quality LoRAs will improve sharpness and detail

• Higher CFG can help lock in his facial features and outfit details more consistently

• Digital Art version: fewer steps gives a cleaner, more refined look; works best alone (400) or with a second character (600), since earlier steps tend to duplicate

• Anime version: fewer steps = cleaner lines, more steps = more faithful to the anime, but with heavier lines and a tendency toward typical anime errors

• Check the prompt examples used in the posted preview images — they show how to structure tags for each outfit

📊 Checkpoints

🎨 Digital Art version

StepsResult 200Improves costume details — already a noticeable improvement over the base ANIMA

⭐ 400Strong focal point for solo Midoriya — recommended for solo generations

⭐ 600Works well for character duos — recommended starting point if not using another LoRA for the second character

800Moderate result

1000Slight improvement over 800

Recommended: 400 alone · 600 with another character

📺 Anime version

All checkpoints produce a recognizable Midoriya — choose based on how clean or how anime-faithful you want the result.

Fewer steps = cleaner lines · More steps = more faithful to the anime, heavier lines and tend toward typical anime errors

⚙️ Generation settings used for preview images

All preview images were generated using:

• Base model: Anima 1.0

• Acceleration LoRA: strength 1.0

• CLIP: anima default

• Steps: 12

• CFG: 1.0

• Sampler: er_sde - beta

👍 One more thing

I'm just getting started with this — if you have a moment, drop a Like and share your generations with the settings you used. It really helps me improve the next versions. Have fun! 🔥