Qwen Multiangle Camera Translate

The Qwen Multiangle Camera node always outputs its camera terms in English - front view, eye-level shot, medium shot. That's fine if your base prompt is English. If it's Chinese, Japanese, or Korean, you've got a problem the model itself creates: it tends to ignore instructions that don't match the language of the surrounding prompt. Slap an English camera phrase on a Chinese prompt and the angle quietly stops working.

Qwen Multiangle Camera Translate is the fix, and it's delightfully boring about it. No LLM, no API call, no translation model - it's a hand-maintained glossary plus a regex. You feed it a prompt, pick a target language, and it rewrites only the camera/shot vocabulary. Everything else - your actual prompt, the <sks> token, punctuation - passes through untouched.

How it works

The glossary lives in camera_glossary.py in the pack and covers exactly the 15 phrases the camera node can emit: the 8 azimuth directions, 4 elevation shots, and 3 distance classes. The translation is a case-insensitive, word-boundary regex substitution - so it won't mangle a word that merely contains "front". Since it only knows camera vocabulary, your creative writing is safe by construction.

Inputs and outputs

It's a two-in, one-out string box, and the tooltips say it all:

• prompt - a multiline string. Link the camera node's Prompt output here, or paste text directly.
• target_language - the dropdown. 中文 (Chinese), 日本語 (Japanese), 한국어 (Korean), or English. English is a pass-through no-op, which is handy for toggling a workflow between languages without deleting the node.
• Prompt (output) - the translated string, wired into your text encoder (or the editor node's prompt) in place of the raw camera output.

Typical chain: Qwen Multiangle Camera → Qwen Multiangle Camera Translate → CLIP Text Encode (or straight into your Qwen-Image-Edit prompt input). The same angle becomes <sks> front view eye-level shot medium shot in English, <sks> 正面视角 平视 中景 in Chinese, or <sks> 정면 아이 레벨 미디엄 샷 in Korean.

Installing it

You almost certainly already have it. This node ships in the same pack as the camera node, so install the pack once via ComfyUI Manager (search "ComfyUI-qwenmultiangle") or:

cd ComfyUI/custom_nodes
git clone https://github.com/jtydhr88/ComfyUI-qwenmultiangle.git

Then restart ComfyUI. There are no extra dependencies and no model files - the glossary is baked into the Python. It's kept in sync with the pack's UI translations, which is why the same terms show up correctly in both places.

Things to know

• It only translates what it knows. If you hand it a custom angle phrase that isn't in the glossary, it sails through unchanged. The source comments note the vocabulary is easy to extend by hand - edit camera_glossary.py to add terms, restart, done.
• Don't translate your whole prompt with it. That's not the job, and it will hand you back mostly-English gibberish for anything outside the camera vocabulary. Feed it only the camera node's output.
• Keep the <sks> token intact - it's the LoRA's activation token and the translator leaves it alone on purpose. If you paste a prompt that lost it, the Multiple-Angles LoRA won't engage.