Layer supports uploading images and assets in several different contexts — as reference images for generation, as training images for custom models, and into Layer Drive for shared workspace storage. This article covers how each works, common issues, and how to fix them.
Where you can upload images in Layer
Where | What it does |
2D Forge | Upload a reference image to guide style, structure, character pose, or composition during generation. |
Prompt Edit | Upload an existing image as the base for editing. The uploaded image is what the model will modify according to your prompt. |
Video generation | Upload a first frame, last frame, or reference image depending on the video model being used. |
Custom model training | Upload a set of training images that the model will learn from. Separate requirements apply here — see the How to Train a Custom Model article for full details. |
Layer Drive | Upload files to the shared workspace drive so they are accessible to all team members across sessions. |
How to upload a reference image for generation
There are three ways to add a reference image in the 2D Forge and Prompt Edit sections:
Click the upload area and select a file from your device.
Drag and drop an image file directly into the reference image box.
Copy and paste an image directly into the reference area using Ctrl+V (Windows) or Cmd+V (Mac). This works for images copied from other applications or from within Layer itself.
If you are using Prompt Edit, you can also trigger it directly from a generated image by clicking the three-dot menu on the image and selecting Prompt Edit. This automatically loads the generated image as your reference without needing to upload separately.
Understanding reference types
When you upload a reference image in the 2D Forge, you can choose how the model uses it. The available reference types depend on the model selected. Common reference types include:
Reference type | What it does |
Reference Image | Uses the image as a general visual reference for the generation. The most broadly supported type across models. |
Style | Guides the visual style and aesthetic of the output, without constraining the structure or content. |
Structure / Outline | Uses the image's composition and structure as a guide, allowing the model to reinterpret the content while keeping the layout. |
Character Pose | Guides the pose and orientation of a character in the output. Only available with character-type models. |
If a reference type appears greyed out or unselectable, it means the currently selected model does not support that reference type. Switch to a compatible model or use the Reference Image type as a fallback, which is the most universally supported option.
Troubleshooting upload issues
"Failed to upload any files" error
This is the most commonly reported upload issue. It appears most often in the 2D section when uploading reference images. Restarting your computer or reloading the page does not always fix it. The most effective steps to try are:
1 | Hard refresh the page using Ctrl+Shift+R (Windows) or Cmd+Shift+R (Mac) to clear the cached version of the app. |
2 | Clear your browser cache for layer.ai specifically, then reload and try again. |
3 | Try a different upload method. If dragging and dropping is failing, try clicking the upload area to select the file manually instead, or use copy and paste. |
4 | Try a different browser or an incognito window. Some browser extensions can interfere with file uploads. |
5 | Check your file size. Very large files, particularly images over 4K resolution, can cause upload failures. Resize the image to 1024x1024 or a similar standard resolution before uploading. |
6 | Start a new session from the Layer home page rather than continuing in the current one, as session state can sometimes block uploads. |
If the error persists after all of the above, contact support with your session link and a description of the file you are trying to upload.
Drag and drop not working
If you can no longer drag reference images into the prompt or reference window, this is usually a session or browser state issue. Try starting a fresh session from the home page. If dragging still does not work, use the click-to-upload method or copy and paste the image directly into the reference area instead.
Reference type options are greyed out
Reference types like Character Pose, Style, or Structure are only available on models that support them. If these options are greyed out, you are either using a model that does not support those reference types, or you are in the wrong section. Ensure you are in the 2D Forge with a compatible model selected. If you are trying to use an image edit model like Gemini Image Edit, you need to use the Prompt Edit section rather than the 2D Generator, as these models are not accessible from the generator view.
Upload is taking a very long time or appears to hang
Large file sizes are the most common cause of slow or stuck uploads. Resize your image before uploading. If the upload appears to have hung, refresh the page and try again with a smaller file. If you are uploading training images and they are not completing auto-captioning, this is also a sign the files may be too large — resize to under 4K resolution before re-uploading.
The wrong files are showing in the file picker
If the file selection window is showing generated images from your session rather than previously uploaded files, this is a known display behaviour rather than an error. You can still navigate to your uploaded files within the picker, or upload a new file directly from your device.
Frequently Asked Questions
What file formats does Layer accept for reference images?
Layer supports common image formats including PNG, JPG, and WEBP. PNG is recommended when working with images that have transparency, as the transparency information will be preserved during generation. For video reference frames, MP4 is the standard format accepted.
Is there a maximum file size for uploads?
There is no fixed published file size limit, but images over 4K resolution consistently cause upload failures and training issues. Resizing to 1024x1024 or a similar standard resolution before uploading is strongly recommended for both reference images and training assets.
Can I use a transparent PNG as a reference image?
Yes. Layer supports transparent PNG files and will preserve the transparency in the output. Using transparent images in your training set or as references is particularly useful when generating assets that need to be composited onto other backgrounds.
Can I paste an image directly from my clipboard?
Yes. You can use Ctrl+V (Windows) or Cmd+V (Mac) to paste an image directly into the reference area from your clipboard. This works for images copied from other applications, websites, or from within Layer itself.
How do I share uploaded files with my whole team?
Files uploaded in a session are accessible within that session. To make files available to all team members across sessions, upload them to Layer Drive from the workspace. From there, any team member can access and use them in their own sessions.
I uploaded an image for training but it did not get auto-captioned. What should I do?
Auto-captioning failures are almost always caused by oversized files. Resize the image to 1024x1024 or similar before re-uploading. If the image still does not receive a caption after resizing, you can add the caption manually in the training interface before starting training.
The reference image box is greyed out and I cannot upload anything. What is wrong?
This usually means you have selected a model inside the 2D Generator that is an image edit model, which can only be used through the Prompt Edit section. Navigate to Prompt Edit directly from the home page and you should be able to upload your reference image there without issue.
Still having trouble uploading an asset or reference image? Contact support with your session link and a description of the file you are trying to upload and we will help.