---
title: "Interactive Rendering Tips"
canonical: "https://rmanwiki-27.pixar.com/space/RU/630915278/Interactive%20Rendering%20Tips"
format: markdown
---
A helpful guide on optimising your interactive render experience with RenderMan XPU
> ℹ️ **Interactive Denoiser**
RenderMan’s amazing batch denoiser runs also in interactive viewport mode, so turning this on will denoise your viewport as you work on your shaders, lighting, camera, and final refinement.
For smaller scenes, such as prop lookdev, you may want to keep the Interactive Denoiser Min Samples at 4; for larger scenes with more geometry, instancing, and shaders, you might want to increase it to 10 or 15.
This tells RenderMan how many samples to render before the denoiser starts its magic.
Altering the Interval can also give you some interactive denoising speedups.
It all depends on your scene and its complexity.
> ℹ️ **Image Resolution**
Reducing image resolution will improve interactive performance, especially in larger, more complex scenes.
> ℹ️ **Scale Resolution**
Scaling the IPR resolution can dramatically speed up interactive responsiveness. For instance, in RfB there is an option to change the resolution scale to 75%, 50%, 33%, 25% and 10%
Solaris has a similar option too.
This, in effect, scales the resolution down and lets the DCC rescale the image back to the viewport size, giving the artist a more ‘blocky’ render, which is often enough to make creative decisions in layout, lighting, and staging.
> ℹ️ **Pruning**
Pruning (or hiding) geometry outside the shot frame or camera frustum will help speed up both RenderMan’s time to first pixel and improve interactivity. You need to bear in mind that objects outside the camera might contribute to the overall look of the shot, such as casting shadows, but if this is not a concern, pruning these objects is good practice.
Pruning objects by distance is also valid.
> ℹ️ **Progressive Pixels**
Setting RenderMan Progressive Pixels to 2 - 4 will increase viewport interactivity speed.
> ℹ️ **Time to First Pixel**
When you first start an XPUGPU or Hybrid interactive render, the NVidia driver needs to do some initial heavy lifting, compiling RenderMan into CUDA code which is suitable for your GFX card.
The second time you restart the IPR session, this work has been cached by the driver so your time to first pixel will be much faster.
> ℹ️ **Camera Distance**
If you have objects with SubDivision applied and are close to the camera, XPU will take longer to start up than if the objects were farther away. A lower MicroPolygon length can also affect XPU’s initial start-up time.
> ℹ️ **Turning off Subdivision & Displacement**
Turning off Subdivision and Displacement will speed up XPU’s time to first pixel, and in many cases, you can still make valid artistic choices without having both features turned on.
> ℹ️ **Increase MicroPolygon Length**
Meshes are diced relative to the camera at the start of the render. Increasing the MicroPolygon length reduces the amount of tessellation work to be done before the renderer starts, thereby speeding up the time to first pixel. This, in turn, results in lower memory consumption.
> ℹ️ **Material Override**
When working on layout or camera staging, you don’t always need to see all the textures and complex shading networks, so you can either apply a simple material override to the scene geometry or use the PxrOcclusion integrator.
Complete your layout work and return to the PxrPathTracer!
Rendering an override material using only the geometry’s DisplayColor can also be used to its a helpful technique.
> ℹ️ **Render Regions**
Never underestimate the power of a crop window or render region of interest to speed up refinement in specific areas of your scene.