Canon has released Dual Pixel 3D Converter, a free Windows application that allows photographers to turn images captured with compatible EOS cameras into three-dimensional models and short animations. The software, also known as DP3D, uses depth information recorded through Canon’s Dual Pixel autofocus technology, potentially giving owners of existing cameras a new way to use photographs without purchasing specialist 3D imaging equipment.
Canon’s Dual Pixel technology divides each sensor pixel into two photodiodes, allowing the camera to compare slightly different views of a subject when calculating focus. DP3D uses this difference to estimate depth and reconstruct a three-dimensional representation of the photographed object. The resulting model can be viewed from different angles, although the quality of the reconstruction depends on the original photograph and the amount of depth information available.
The software works with supported Dual Pixel RAW files from compatible EOS cameras, including older models such as the EOS R5 and EOS R6 Mark II. Photographers must enable Dual Pixel RAW recording before shooting and use a compatible RF lens, as the application does not currently support every lens or camera configuration. Canon recommends photographing subjects with clear separation from their backgrounds, avoiding excessively complex shapes and using suitable exposure settings to obtain more reliable results.
Once the images have been transferred to a Windows 11 computer, they can be imported into DP3D for processing. The application allows users to adjust the apparent depth, isolate subjects and change backgrounds before exporting their work. Finished models can be saved in OBJ or GLB formats for further editing in applications such as Blender, while a separate preview function produces short animations showing the reconstructed subject from different viewpoints.
There are limitations to the technology. Because DP3D estimates three-dimensional shape from the information captured in a single photograph, it cannot reproduce every surface or hidden detail accurately. Subjects with complicated outlines or substantial differences in depth may therefore produce less convincing results. The software is likely to be most useful for relatively simple product photographs and other controlled studio subjects rather than complex scenes.
Canon Dual Pixel 3D Converter is currently available free of charge for Windows 11, although Canon states that its terms and availability may change. Photographers can download the application and sample images from the official Canon DP3D website.
Product description
Dual Pixel 3D | Create a 3D Model from a One Shot
Dual Pixel 3D Gallery
3D Photo – Convert the entire image, including the background, into a 3D model. Shape, thickness, depth, and overlapping elements. Explore details that a photo alone cannot fully reveal, from perspectives unique to 3D.
3D Models – Automatically isolate the main subject in an image and generate a highly detailed, textured 3D model.
Expand your visual expression with 3D models – Use the extracted 3D model in 3D editing software such as Blender or in PowerPoint. Create 3D videos with motion, viewpoints, and presentation tailored to your vision.
Technology – Dual Pixel 3D is a 3D model generation technology based on Canon’s proprietary on-sensor phase-detection autofocus technology, Dual Pixel CMOS AF. Using Canon’s Dual Pixel 3D Converter software for PCs, you can generate a highly detailed 3D model of the area captured in a one shot from image data* recorded with a compatible camera and lens in Dual Pixel RAW (DPRAW) or JPEGs with distance data.
01. Capture – DPRAW/JPEG Images Captured Using Dual Pixel CMOS AF. Each pixel in the Dual Pixel CMOS sensor performs both imaging and phase-detection AF functions. With two photodiodes in every pixel, the sensor captures phase-difference information across all pixels, enabling it to estimate the distance to the subject.
02. Convert depth information into 3D geometry – Based on the phase-detection data obtained from Dual Pixel CMOS AF, it generates a 3D model using an exclusive algorithm. Using the high-quality photographic data for the 3D model surface texture captured with EOS, it can produce realistic 3D models.
03. Generate a 3D model – OBJ / GLB. The area visible at the time of shooting is generated as a 3D model with high-quality photographic textures. The generated 3D model can be exported in OBJ or GLB format.
Photography Guide
Shooting Tips for Better 3D Model Results
When generating a 3D model with a convincing sense of depth, it is important to photograph the subject under conditions that make its shape easy to recognize. By following three sets of recommendations—Camera Settings / Shooting Settings, Subject Texture and Shape, and Composition, Angle, and Distance—you can generate more accurate 3D models.
If the subject is dark or the details are unclear, use a flash or other lighting source, and shoot at the lowest possible ISO to minimize noise. 3D models cannot be generated at ISO values greater than 3200.
Subject texture and shape (Recommended)
Items with a matte texture
- Ceramic and wooden items
- Food and leaves with little sheen or gloss
- Ornaments photographed to minimize reflections
Objects with thick, relatively simple shapes
- Shallow plates
- Food items with rounded shapes
- Flowers with round petals, or flowers with narrow petals but rounded outlines
- Houseplants with large leaves
Characteristics of backgrounds that are easy to separate
- A solid-color background
- There is a color contrast between the subject and the background
- No shadows on the subject caused by artificial lighting or natural light
Composition, angle, and distance (Recommended)
Center the subject with space around it. Position the subject in the center of the frame and compose the shot so that the main subject is clearly defined. The dark gray areas indicate blind spots. Shooting from a higher angle and farther away can help reduce blind spots.
Shoot from a slightly elevated angle and farther away to reduce blind spots. Blind spots refer to areas that are not visible during shooting. Dual Pixel 3D can only convert the area visible in the frame at the time of shooting into a 3D model. Therefore, areas outside the field of view—which are blind spots—cannot be converted into a 3D model, and the subject may appear stretched or distorted. Reducing blind spots during shooting makes it easier to generate natural-looking 3D models with minimal distortion.
Subject Distance: 5-10 Times the Focal Length. A distance of 5–10 times the focal length is recommended between the camera’s sensor plane and the subject. For example, with a 50mm lens, aim for a distance of approximately 25 to 50 cm.