Nodes/ComfyUI-Qwen-FrameKit/Camera Angle Prompt Combine
ComfyUI Node

Camera Angle Prompt Combine

The two-line glue that keeps your camera prompt out of your edit prompt

By uron83·Created 13 days ago·Updated 7 days ago· 0
Camera Angle Prompt Combine
    • prompt
    camera_prompt
    edit_promptEdit the image while preserving identity, pose, and composition.

    This is the boring node in the Qwen FrameKit pack, and boring is the point. Qwen-Image-Edit's camera-angle LoRAs are triggered by a token like <sks> followed by a camera description - <sks> front view eye-level shot medium shot - and that token string has to sit right above your actual edit instruction in the prompt. Type that by hand into a giant text box and you will, eventually, mangle it. This node exists so you never have to.

    What it does

    CameraAnglePromptCombine takes two strings - a camera_prompt and an edit_prompt - and joins them:

    <camera_prompt>
    <edit_prompt>
    

    That's the whole job. It strips surrounding whitespace, puts a newline between them, and returns one combined prompt. If either side is empty it just returns the other, so it degrades gracefully mid-workflow.

    Why you'd actually wire it in

    Two reasons. First, the FrameKit angle nodes generate the camera prompt for you - QwenImageEdit2511AngleCamera emits it from its azimuth/elevation/distance sliders, and the legacy CameraAngleSelector emits a list of them. Combining keeps that generated text as a separate moving part instead of buried in an edit prompt you're always tweaking. Second, the default edit_prompt - "Edit the image while preserving identity, pose, and composition." - is genuinely a good Qwen-style instruction. Leaving it in place is the easy way to keep the edit model from drifting the face or the framing while it rotates the camera, which is this ecosystem's most common failure.

    Wire the prompt output into the CLIP Text Encode that feeds your Qwen-Edit sampler, same as you would a hand-written prompt.

    The honest gotcha

    It doesn't validate anything. Paste the wrong trigger token and it concatenates happily, and Qwen will cheerfully ignore a trigger it doesn't recognize. Keep the camera trigger on its own line (which this node does for you), and if your angles are coming out wrong, fix them upstream in the angle node, not here. There's no magic in the combine step - it's string concatenation with good defaults.

    Installing it

    It ships in the same pack as the angle nodes, so installing ComfyUI-Qwen-FrameKit brings it along:

    cd ComfyUI\custom_nodes
    git clone https://github.com/uron83/ComfyUI-Qwen-FrameKit.git
    

    or use ComfyUI Manager and search "Qwen FrameKit". Restart ComfyUI and hard-refresh the browser. No extra Python packages - the pack's whole dependency story is "none". If you've already installed the pack for the angle-camera node, this one's already on your node list under camera → Camera Angle Prompt Combine.

    Categorycamera

    Inputs (2)

    NameTypeDefaultDescription
    camera_promptSTRING
    edit_promptSTRINGEdit the image while preserving identity, pose, and composition.

    Outputs (1)

    NameTypeDescription
    promptSTRING