INTRO
A short – need to know – route through using Reality Capture to process a set of photos to create and export a 3D model which can then be used for 3D printing, as a model in CAD, for a render subject etc.
Reality Capture is effectively free for small projects and downloadable from Epic Games – and is installed on all PCs in design studio computer areas.
START REALITY CAPTURE
Fire up Reality Capture from the Windows start menu – you can skip any login prompt. Choose the “1+1” layout from the top left selector.

IMPORT
Next step is to locate all your photos in the Windows file browser and simply drag them all into the RC Window. Alternatively click on the INPUTS button and browse to the images.
For optimum results images should be:
- Sharp: in focus, no motion blurring, etc.
- Well lit: so all the features of the subject are visible (and not in black shadow)
- Overlapping : ideally 60% or more for adjacent images so that the software can tie common points together
It can be useful to go through the photos first to eliminate unsuitable ones and / or alter brightnesses to show details better.
Note – you can also drag a movie into here, and RC will extract frames from the movie and use those.
Once the images have come in they can be seen as a list in the folder structure on the left hand side.
There is a START button which is a short cut for doing all the steps necessary in one go. This shortcut might work but often it is necessary to go through each step in turn to have a more controlled operation. Depending on the machine spec and number of photos this can take some time (RC will give an estimate for completion time).

If this one-button shortcut DOES work then the model can be exported without going through the next steps – in which case skip to the EXPORT section below, or CROP MODEL if you needs to cut the part down.
ALIGNMENT
This step will do a quick parse of the photos where it will try to detect common points in the photos and tie these together. It will give a rough idea of how well the model will turn out.
Click on the Alignment tab at the top and then the Align Images button

This will ideally give a single “sparse” point cloud of the model. Reality Capture will recognise common geometries from the photographs as components – and how many of the photos it recognises those components from.
Note:
- The best scenario is if the photos are recognised as a single component, using all the photographs
- If a component has not used all the photos it is still possible to take it forward
- IF RC has created many components then the options are to pick the one with the most camera matches (or if another looks better select that), then take that component forward
- OR, it is possible to align components together manually – but this can be a difficult process which involves finding common points on different components and forcing a realignment. It is a LOT easier if the photos are such that this step is not necessary.
MESH / TEXTURE MODEL
With the sparse point cloud component selected the next step is to create a solid mesh of it via the MESH MODEL tab. This step can take a while – and there is an option for creating a quicker preview of the mesh if necessary, otherwise use the regular NORMAL DETAIL button to create a mesh. (There is also a HIGH DETAIL option but it is not usually necessary to generate such a high poly mesh).

This step will give a solid, flat single colour model . To have a better view of the model you can give each vertex in the model the right colour; click on the COLORIZE button. You can do a full texture map by clicking on the TEXTURE button at this point too – but if there is any cropping of the model necessary it is best to do the texture step after that.
CROP MODEL
If there are areas at the edges of your model that you do not want, or RC has auto filled and “boxed” out the edges then you can crop these areas out altogether and just focus on the important area of the model. There are many mesh editing tools in RC but the easiest is to pull a box around the area and snip everything else out. This is a really useful step:
First got to the far right TOOLS tab at the top, them pick the SET RECONSTRUCTION REGION button

This will show the draggable handles of the box as coloured points – pull these around so just the area of interest of the model is enclosed, and / or drag the curved rotate widget to rotate the box. (if you hop into the VIEW tab there is an option there to view in plan / elevation which can make selection easier).
Once the crop box is in place then click the CUT BY BOX button over on the right hand side. This will bring up an option (on the left side) to CUT OUTER, which will cut away anything outside the box – and importantly fill and close the model. This is a better point to carry out the TEXTURE operation from the MESH tab if required.
EXPORT
Once the model is cropped and textured then final step is to export this model from RC for use in other applications. Go to the MESH MODEL tab and look for the button LEVELS OF DETAIL.

Clicking this will give the option to choose a file type. FBX is a common format that will work with most modern 3d applications (inc. Rhino, Blender, Maya etc). For older softwares – and often for 3d printing, or laser cut slicing software -you may need to use the OBJ format.
In either case the LOD (“Level of Detail”) option will allow you to create a number of exports: in the MODEL COUNT field select 3. This will export 3 versions of the model at different levels of complexity – smaller models will work more quickly and efficiently than larger models. Larger models will have more detail but can be more demanding when used in 3D modelling software (especially if it is copied around, edited etc.) This LOD option is very useful for being able to preview a range of options before settling on a final one to take forward.
Once this operation is complete then 3 FBX or OBJ files will be in the folder you specified. These are now ready to be utilised in modelling or 3d printing or any other process you need them for.