Silence_Mix

I used to like V2 the most, but now I like V4 the most.

V4 is here! Since I really enjoy this series, let's analyze what this version brings.

First, let's briefly define the two styles I mentioned:

Anime Style: This resembles characters whose skin appears to glow (not the greasy shine generated by adding the "shiny" prompt). The material of the clothing significantly differs from real-life fabric.

Flat Shading Style: This resembles characters whose skin doesn't obviously shine. The clothing material is closer to real fabric, and the shading is uniform.

In this version (V4), whether using built-in characters or anime character LoRAs, the presentation leans heavily toward anime style with almost no realistic style.

However, the backgrounds, compared to previous versions, are less layered, less realistic—essentially flat shading style. For example, the clouds in the sky: in earlier versions, they appeared as numerous white dot patterns, but in this version, they look like clusters of white.

So, depending on your preferences:

1.If you want to:

Use built-in characters

Use anime character LoRAs

Prefer characters in anime style

Prefer realistic backgrounds Choose V1

2.If you want to:

Use built-in characters

Use anime character LoRAs

Prefer characters in anime style

Prefer flat shading style backgrounds Choose V4

3.If you want to:

Use built-in characters

Use anime character LoRAs

Prefer characters in flat shading style

Prefer flat shading style backgrounds Choose V3

4.If you want to:

Use built-in characters

Do not use anime character LoRAs

Prefer characters in anime style

Prefer more realistic and stable backgrounds Choose V2

Since version 3 has been released, I am once again testing the pros and cons of each version. However, before doing so, I want to define the anime style and flat coloring that I refer to.

The anime style is more like the characters' skin will shine (not the greasy shine generated by adding the 'shiny' prompt), and the clothing material is noticeably different from real clothes.

Flat coloring style is more like the characters' skin will not significantly shine, the clothing material is closer to real clothes, and the colors of the characters' skin and clothes are uniform (I guess this is what the mode author refers to).

So:

V1: Good background, when using anime character lora and built-in characters, it generally produces anime style, rarely appearing 3D or realistic style.

V2: Good background, stable background, when using anime character lora, it occasionally produces a realistic style close to real people.

V3: Good background, background color richness is reduced (flat coloring style), when using anime character lora and built-in characters, it rarely appears 3D or realistic style, but the characters tend to be in flat coloring style rather than anime style.

In summary, for those who like flat coloring style with a good background, choose V3. For those who like anime style and use built-in characters, choose V2. For those who like anime style and use anime character lora, choose V1.

By far, best H Checkpoint i tested, especially if you like dogs and have fur fetish, i used interspecies combinations and this have alot artists to use without a single Lora!!!

Tried V6 and LORAs completely break it and add tons of artifacts. Without any character/pose/style lora it works perfectly. If creator can solve the LORA compatibility issues this can be best model on the site

I really like this model, so I’m testing the new version.

About V6:

Background: As long as there’s no light source (for example: no sun or nighttime lamps), the background remains as spectacular as ever—just like the initial V1 version—with near-realistic illustration–level detail. If there is a light source in the scene, the image slightly loses that realistic illustration feel. Additionally, V6 backgrounds sometimes include light particles and tend toward a blurry background.

Characters: Hair fully leans toward the modern anime glossy-paint look, and even with a hair-detail LoRA added, it’s hard to capture the texture of each individual hair strand. Skin uses a flat-shading style (see earlier comments for a description of this style), similar to V5 (and V3).

Character LoRA: Surprisingly, it’s highly compatible—back to V1 levels.

Conclusion: You need high character-LoRA compatibility, realistic backgrounds, and the option for Flat Shading Style, so V6 is your best choice.

Under negative-prompt conditions, you may need to add light particles to reduce unwanted light spots and introduce a blurry background to restore realism.

In any case, thanks to the model’s author.

V5 is essentially an enhanced version of V3’s flat coloring style, though the background and hair occasionally feature a glossy painting technique reminiscent of modern animation (for example, the black in the hair isn’t a pure, uniform black but appears as if it’s been coated with glossy lacquer).

Additionally, the effect of the anime character LoRA seems somewhat weaker.

Beautiful model, thank you for making & sharing this model❤️