Uploaded is not model-compatible
A stored file can still use an unsupported format, size, duration, dimension, media type, or input combination.
Papiboom Studio
GUIDE / REFERENCE MEDIA
Trace the reference path from upload and media readability to model compatibility, selected node output, and prompt direction.
No subscription required / Pay-as-you-go credits
These synthetic examples were checked against Papiboom's local request validators on October 12, 2026. A pass means this input-contract check passes, not that upload, provider moderation, generation or output quality has been tested. No generation output or acceptance rate is claimed.
Happy Horse 1.0 and 1.1 have different input contracts. Do not infer video support from the family name. For a new project, start with the reference-led test brief; compare Seedance versions before choosing. Fewer restrictions concerns filtering, not these input requirements.
/ Papiboom editorial
An image can appear in a project and still fail at generation. Upload storage, media decoding, node selection, reference role, model limits, and prompt direction are separate parts of the path.
A reference image may not work because the file did not finish uploading, cannot be decoded, is connected from the wrong node output, uses a role the selected model does not support, exceeds that model's limits, or reaches the model without clear direction. Seeing a thumbnail only confirms one part of that path.
A stored file can still use an unsupported format, size, duration, dimension, media type, or input combination.
A style image, first frame, motion video, and audio source travel through different model controls.
When a connected node has several outputs, Papiboom sends the currently selected history result downstream.
The visible symptom narrows the failing layer. Fix that layer before changing every prompt, model, and source at once.
The workspace has not received a usable source
Check connection, size, file type, and visible upload progressThe upload exists but cannot be decoded or materialized
Open the asset directly and try a standard exportThe selected model does not accept that role or combination
Check media type, count, frame mode, duration, and required inputsThe request reached generation but the upstream task rejected or failed
Preserve the task ID and read the specific failure reasonThe media arrived but its intended role was unclear or weak
Select the intended output and describe the source role explicitlyThe same file can be valid in one role and invalid in another. Decide whether it should define appearance, frame boundaries, motion, or sound.
Use a model that accepts images; state what should remain consistent
Treating the image as a guaranteed pixel-for-pixel copyAdd the first frame before the last; keep frame mode separate from mixed-reference mode
Mixing frame mode with a separate multimodal-reference setUse a model that accepts video and keep the source within its duration and size limits
Sending video to a text-or-image-only modelUse an audio-capable model; some models also require a visual source
Adding audio without the visual source required by the modelThese limits reflect the current Papiboom integrations as reviewed on October 12, 2026. Model contracts can evolve, so the live model card remains the final check before submission.
Up to 30 images, 10 videos, and 10 audio files, or first + last frame
Frame mode requires adaptive aspect and first before last; do not mix with general referencesUp to 9 images, 3 videos, and 3 audio files, or first + last frame
Last frame requires first; frame and mixed-reference modes are separateUp to 9 images, 3 videos, and 3 audio files, or first + last frame
Audio needs a visual source; video and audio duration totals are limitedExactly 1 character image + 1 motion video
Duration follows the usable motion clip; this is not a general mixed-reference modeUp to 9 image references
No video, audio, or first/last-frame input in the current integrationUp to 1 image reference
No video, audio, or first/last-frame input in the current integrationUse one known-good source first, then add complexity one item at a time. This preserves a working baseline and reveals the exact incompatible input.
Confirm the image, video, or audio plays outside the node and has finished uploading.
Decide whether it is a general reference, first or last frame, motion source, or audio source.
Check that the selected model accepts that media type, role, count, duration, and combination.
Begin with a single authorized reference and confirm that its capsule appears on the generation node.
If the source node has history, click the exact image, video, or audio output you want to send.
A connected source is submitted without an @ token, but @ can clarify which source controls subject, style, action, or framing.
Reconnect the remaining media one at a time so count, duration, or mode conflicts are visible.
Preserve the exact message and task ID when the request reaches generation but still fails.
Standard formats reduce browser and provider decoding surprises. They do not override a model's own size, duration, dimension, or content limits.
Use broadly supported exports before investigating a model-specific problem.
Every generation model can apply narrower constraints.
Short answers for references that upload, connect, or generate differently than expected.
A successful upload only proves that the workspace received the file. The selected model must also accept that media type, reference role, file properties, and combination of inputs.
In Papiboom, a connected reference is submitted even when no @ token appears in the prompt. An @ reference token is useful when you want to state clearly which source controls the subject, style, action, or another part of the scene.
Only when the selected model accepts video references. Motion Control is a separate input contract that requires exactly one character image and one motion video rather than a general multimodal reference set.
Models with frame control require a first frame before a last frame. In the current Papiboom integrations, frame mode is also kept separate from multimodal-reference mode.
Papiboom sends the currently selected node-history result to a connected downstream node. Select the intended image, video, or audio result before running the next generation.
JPG, PNG, or WebP are practical image choices; standard MP4 with H.264 is a practical video choice; and MP3 or WAV are practical audio choices. Model-specific size, duration, and dimension limits still apply.
If the media path is healthy, continue with the broader failure checklist or compare the live input contracts before moving the same sources to another node.
Bring the prompt and authorized image, video, or audio reference to Papiboom. Connect each source visibly, choose a compatible model, and retry without rebuilding the workflow. Model capabilities, Papiboom rules, and applicable law still apply.