I think it's time to end the updates for this series. I've almost run out of things to modify, except for some minor adjustments to a few less-than-ideal models, but that doesn't seem worth the effort.

Recently, I'll try to research the Z-IMG model and see if I can perform some fine-tuning or LoRA training on it.

Each of the three versions has distinct features, but the overall changes are minimal.

You can see their main styles through the intuitive preview images:

v1

• Version 1 has a strong cartoon style, but the overall image appears hazy.

v1.3

• It's a minor tweak to version 1.0, with filter-related effects removed.

V2

• Based on Version 1, Version 2 reduces the proportion of the cartoon-style filter, adds an extra anime style while keeping the overall tone unchanged, and features a richer color distribution.

v2.1

• Based on Version 2, Version 2.1 adjusts the character's body shape, skin texture, and lighting. It leans more towards a realistic style, with the cartoon style significantly weakened.

v2.2

• Based on version 2.1, the main adjustments are to the visual effects. Version 2.2's advantage lies in its ability to output images with a near-realistic style but anime-like composition.

• Highlight: Highly compatible with various character types in Lora.

v2.3

• not good enough

v2.4

• This version incorporates some previous LoRa styles to achieve better environment, lighting, facial effects, and body effects. Of course, it will definitely have some bugs, but I haven't found any yet, so I've marked it as unstable.

• Version 2.4 fixed the issues in the unstable version and was re-uploaded, but the unstable version still has some unique features, so I kept it.