Reference Images

There are different guidelines for use of Reference Images depending on the tool, please see examples below.

Lifestyle Photography & Graphics and Placement Prints

Use reference images sparingly to get unique results. Since Raspberry AI is a visual platform, it will adhere closely to images. Use the similarity slider to give room for more creativity. See example below where the user is attempting to create a similar image featuring a black leather bomber jacket, instead of the dark green nylon bomber jacket in the Reference Image.

prompt example

A young woman wearing baggy cargo jeans, a black leather bomber jacket, black heels, and glasses.

Reference Image:

Failed outputs when Similarity Slider is set to 50%:

Successful outputs where Similarity Slider is set to 25%:

Product Photography

Reference Images used in the Product Photography tool should have a clean background, and no model present in the image. See examples below:

This image has a model and imagery in the background. This reference image will not generate a Product Image.

Yes!

For best results, your Reference Image should also be a product image, which is a flat lay or ghost mannequin image with no model present on a neutral background.

Prints and Patterns

Please reference Prints & Patterns Help Doc for comprehensive information on how to utilize Reference Images with this tool.

Sketch to Render

Reference sketches used in the Sketch to Render tool should be a clearly detailed sketch, created using Adobe Illustrator, pen, pencil, marker or otherwise. For best results it should be a file with a clean white background, or if the image is scanned from a sketch on paper, the paper should not have any grid lines or other marks as they will show up in the final render. See examples below:

The sketch still has visible croquis lines that were not fully erased, which will show up in the final image.

The sketch is clean, but on grid paper. The grid lines will show up in the final image.

Yes!

Your Reference Sketch should be clean, and on a white or neutral background.

3D Avatar to Photorealism

Yes!

The 3D Avatar to Photoreal tool works best on 3D avatars with facial features, skin, and hair. It is designed to work with digital avatars created by companies such as CLO, Browzwear, Daz 3D, Style 3D and more! Alvaforms with realistic skin and hair will also work. If your garment features graphics, they may be slightly altered - if this is the case you can use the Edit tool to correct.

Faceless avatars: The tool can only enhance what already exists. If your avatar is faceless, the output will not have facial features. If your avatar is bald, the output will also likely be bald.

Alvaforms with canvas “skin”: The tool will not work on dress form style avatars covered in a canvas material, as there are no facial features.

Ghost avatars: The tool will not work on ghost avatars, where the render is a hollow body with no avatar visible.

2D Sketch

The tool will generate a sketch of the image and will include all details, including fabric wrinkles and folds, and prints. For best results, reference images should be a clean image of the product with no person and a clear background. If generating a 2D Sketch of apparel, the garment should be flat, with as few creases as possible.

This image has a model and a busy background, so the resulting sketch will be the same and not highlight the garment.

The bag is not laid flat, and is featured on a chair. There are a lot of shadows along the wall. All details will show up in the final sketch.

Yes!

For best results, your Reference Image should be a Product Image which is a garment or accessory laid flat on a neutral background, with as few wrinkles and creases as possible.

Background Generator

Your Reference Image should have a model or person with a neutral, clear background (example: studio backdrop, solid color background). Please reference Background Generator Help Doc for comprehensive information on how to utilize Reference Images with this tool.

The Describe Image feature will pop up automatically when you enter a Reference Image. This feature allows you to precisely describe whatever image you enter, and provides a description that can be used as a Prompt. You can use the Prompt as is, or change and tweak details to get the description closer to your vision. Regenerate for new descriptions.