# Easier Poses Ref2V MM-H3 - Semi Transparent Canny — retained record

> Retained reddit record from reddit:StableDiffusion.

Canonical: https://highsignal.app/data/records/33214c81280348db

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:StableDiffusion)  published 2026-11-02 02:48 ingested 2026-11-03 02:40

[original source ↗](https://www.reddit.com/r/StableDiffusion/comments/1xllwpg/easier_poses_ref2v_mmh3_semi_transparent_canny/)

## retained text



```
Disclaimer: With Prompting, Normal images WORK (shown in the presented video), with this method " I " have found it to work with less bleeding and more variety of outputs.

When working with references sometimes the reference image of a pose ends up bleeding into the video, replacing environment, clothes, or colors. Then i tried canny, and it works, but it has a problem, when the pose is too unusual or complex, H3 tends to ignore or hallucinate many things. so i found that with the og image at a low opacity, the model understand that it's a canny and what the elements in it are.

can be useful on videos, but that's outside my scope of use. so idk what tools are for canny videos.

To end, why to use it? : 1. less likely to bleed into the video. 2. More variety changing the prompt (less strict) 3. very useful for complex/new positions that H3 may not know.

Steps: 6 | Lora: [Dmad-Larry dareties mix](https://civitai.red/models/2837571/minimax-h3-turbo-loras?modelVersionId=3391964) | MP: 0.7  
Prompt:  
\[Visual Style\]

Dynamic natural shot. the video cuts multiple times. many shot cuts

\[Global Constraint\]:

overall\_soundscape: N/A

non\_diegetic\_music: N/A

\[Asset Mapping\]

<Subject 1> is the woman her appearance comes from <Picture 1>, face, hair, clothes

<Subject 2> is the man his appearance comes from <Picture 1>, face, hair, clothes

<Picture 1> is the reference of <Subject 1> and <Subject 2> and location.

<Picture 2> is the reference pose and angle shot composition at \[Shot 2\]

<Picture 3> is the reference pose and angle shot composition at \[Shot 3\]

<Picture 4> is the reference pose and angle shot composition at \[Shot 4\]

\[Chronological Actions\]

\[Shot 1\] The video opens with an exact mirror of <Picture 1>

\[Shot 2\] at 01.000,00,  the shot cuts, in the same location as in <Picture 1>, now the woman sits on the floor,  hands on the floor, legs raised, the man stands, his feet on her shoulders, bending grabbing her feet.

\[Shot 3\] at 02.500,00,  the shot cuts,same location, now <Subject 2> is on the floor, his legs raised. <Subject 1> is balancing her body on his feet, her stomach on his feet.

\[Shot 4\] at 04.000,00, the shot cuts, same location, now the man Kneels on the floor in a child's pose with his torso folded forward and arms extended straight out in front,the woman Kneels on top of the man's back in a curled, folded posture with her hands clasped behind her back.
```



document fetched 2026-11-03 00:32 · canonical https://reddit.com/r/StableDiffusion/comments/1xllwpg/easier_poses_ref2v_mmh3_semi_transparent_canny
