developerjeremylive/Qwen-Image-2.1-Multiple-Angles-LoRA-etheroi

Qwen-Image-2.1 · Multiple-Angles camera-control LoRA

Reshoot any object from a controlled camera position: feed a photo, ask for a different azimuth/elevation, get the same subject from that angle. First camera-control LoRA for Qwen-Image-2.1 — the newest Qwen image base (unified generate + edit, native RGBA), where this format did not exist yet.

• Trigger: <mva> — every caption/prompt starts with it
• 36 poses: 12 azimuths (30° steps) × 4 elevations (0° eye-level / 30° elevated / 60° high-angle / 90° top-down), plus a close-up variant of each → 72 addressable framings
• Recommended checkpoint: step 1,000 (checkpoints/…_000001000.safetensors) — angle-following, identity preservation and detail peak together there; step 1,500 is the alternate, step 2,500 bakes harder (drifts toward copying the reference view)
• Strength: 0.8–1.0 in ComfyUI / diffusers

Real photos

In-distribution examples are above — but does it work on everyday photos it never saw? Dog, cat and apple from Wikimedia Commons, step-1,000 checkpoint, same prompt grammar:

360° orbit — one input, all 12 azimuths

Twelve separate generations (one per 30° azimuth step, eye-level), looped — the camera circles the subject while the cat, cushion and scene hold together:

Individual pose comparisons

Golden retriever · <mva> front-right quarter view, eye-level shot

Golden retriever · <mva> back view, eye-level shot

Cat · <mva> front-right quarter view, eye-level shot

Cat · <mva> back view, eye-level shot

Apple · <mva> front-right quarter view, eye-level shot

Apple · <mva> back view, eye-level shot

Generalizes to real photos: the subject rotates to the requested view and the scene mostly survives, though identity fidelity on out-of-distribution subjects is good-not-perfect (fine fur detail softens). Photo credits (CC): Golden Retriever Carlos, Cat03, Red Apple.

Object examples (training distribution)

Front view in, captioned angle out — rendered at the recommended step-1,000 checkpoint:

<mva> right side view, high-angle shot

<mva> back view, eye-level shot — the front features are gone; genuinely the back of the object

<mva> front-left quarter view, eye-level shot close-up

<mva> left side view, eye-level shot

What the base model does with the same captions

These are step-0 renders from training — the adapter is zero-initialized, so they show the base Qwen-Image-2.1 attempting the identical caption + control image: the camera mostly stays on the original front/three-quarter vantage in both natural-language and <mva> phrasings.

base · <mva> right side view, high-angle shot

base · <mva> back view, eye-level shot

base · <mva> front-left quarter view, eye-level shot close-up

All 40 sample grids across training (4 poses × every 250 steps, fixed prompts) are in samples/ — including the step-2,500 grid if you prefer the stronger-baked look.

Prompt format

<mva> {azimuth}, {elevation}[ close-up]

Azimuth (12 steps of 30°)

Label	Azimuth
front view	0°
front-right view	30°
front-right quarter view	60°
right side view	90°
back-right quarter view	120°
back-right view	150°
back view	180°
back-left view	210°
back-left quarter view	240°
left side view	270°
front-left quarter view	300°
front-left view	330°

Elevation

Label	Elevation
eye-level shot	0°
elevated shot	30°
high-angle shot	60°
top-down shot	90°

Append close-up for the 62%-crop framing. Example:

<mva> right side view, high-angle shot close-up

Usage

Gradio Workflow Space — akhaliq/qwen21-multiple-angles-workflow: canvas UI, photo in → camera knobs out.

ComfyUI — load the LoRA with any Qwen-Image-2.1 workflow (LoRA strength 0.8–1.0), pass your photo as the edit/reference image, prompt with the format above.

diffusers (QwenImage21Pipeline, edit mode = pass the image):

image = pipe(
    prompt="<mva> back view, eye-level shot",
    image=input_image,          # front view of your subject
    num_inference_steps=40,
    true_cfg_scale=1.0,
).images[0]

Training

• Data: 5,028 angle-labeled edit pairs from zeyuanyin/Dome-Objaverse (CC-BY-4.0) — 2,514 objects, front view → 2 target views each, composited on a studio background. Published: akhaliq/qwen21-multiple-angles-train-v1.
• Trainer: ai-toolkit, arch: qwen_image_2, control-image conditioning (block-causal reference attention), rank 32 / alpha 32, adamw8bit, lr 1e-4, weighted timesteps, fp8 (DiT + TE), caption dropout 0.05, 512², 2,500 steps.
• Checkpoints: all ten 250-step saves + samples are in checkpoints/ and samples/ — step 1,000 recommended, 1,500 alternate, 2,500 stronger-baked.

Scope & limits

• Trained on Objaverse objects; real photos generalize (see Real photos above) but identity fidelity on fur/fine detail is good-not-perfect.
• Trained at 512²; larger canvases inherit the style but were not trained on.
• Angles interpolate reasonably between the trained 30° steps; exact azimuth/elevation numbers in prompts are not parsed — use the labels above.

Credit to fal's Multiple-Angles LoRA for Edit-2511 for the caption grammar this format standardized on.