Summary
This article teaches you how to create high-quality renderings in Corona for 3ds Max 6x faster than your typical workflow. Learn how to use the ‘Save 50%’ feature and Intel AI Denoiser to get a result in 7 minutes that would normally take 41 minutes.
Table of contents
- What are you going to learn?
- Why does speed matter in the archviz industry?
- How do you use the downscaling method in Corona?
- Watch the full video tutorial
- How do downscaled renders compare to traditional rendering?
- When should you use the downscaling workflow in Corona?
- Best use cases for downscaling
- What are some of the trade-offs of the downscaling method?
- Frequently asked questions
What are you going to learn?
You’re going to learn how to use a downscaling workflow in Corona to achieve photorealistic results 6x faster than traditional methods.
High-quality renderings often take a long time to finish, but you can significantly cut that time by applying this small trick in your workflow: you first render your image at twice the final resolution, then use the Intel AI Denoiser and the 'Save 50%' VFB option to shrink it down to its final size.
By the end of this article, you will know how to:
- Set the optimal render resolution for downscaling
- Configure the Intel AI Denoiser at the perfect setting for crisp results
- Use the new ‘Save 50%' feature in the Corona VFB for instant resizing
- Adjust the ‘Image Filter’ to ‘Tent’ to preserve fine details
Chaos Corona rendering of a kitchen
Chaos 3D Team
Why does speed matter in the archviz industry?
Faster rendering times allow archviz artists to meet deadlines and iterate on designs more freely without locking up the hardware for hours.
In architectural visualization, time is often the most precious resource, and long render times can quickly become a bottleneck, stopping you from experimenting with different lighting or materials.
With this downscaling method, you solve the following pain points:
- Speed: Deliver final-looking images to clients faster compared to traditional rendering.
- Creative freedom: Test multiple styles and options in the same amount of time that you’d normally take to create one image.
- Hardware limitations: Skip the overnight renders and have your computer ready to go again in minutes, not hours.
How do you use the downscaling method in Corona?
In Corona, you use the downscaling method by rendering at twice the size of your desired resolution, applying the Intel AI Denoiser at a specific threshold, and using the new ‘Save 50%’ feature in the Virtual Frame Buffer (VFB) to instantly resize the image.
While this technique has gained massive popularity recently, much of the credit goes to Render Ram, who popularized this specific downscaling workflow for the Corona community.
Here’s a step-by-step overview of how to do it.
1. Set your render resolution
Start by setting your render resolution at twice the size of your desired resolution. For example, if you need a full HD image, set the resolution to 4K. This provides the extra pixel data needed to hide noise during the downscaling.
2. Configure your Intel AI Denoiser
The next step is key: in your Render Setup window, select the Scene menu and under Denoising, choose Intel AI Denoiser. Here, you will notice the default setting for the Denoiser amount is 0.65; change that to 0.85 for this method.
Pro tip: The 0.85 denoiser amount is the sweet spot for preserving the finest detail and removing excessive noise. 😉
3. Increase the Noise Limit threshold
Because downscaling naturally hides small artifacts and grain, you don’t need to wait for the perfect render, allowing you to push the Noise Limit threshold. For this method, set the Noise Limit anywhere from 20% to 30% — this lets the software finish the calculations much faster!
4. Switch the Image Filter to Tent
In the Render Setup, under System menu, look for the Image Filter setting and switch it to Tent. This helps preserve fine detail and gives the best overall result. You’re ready to start rendering now.
Pro tip: Adjusting your Image Filter to 'Tent' is a critical step in this workflow to ensure fine details remain crisp after downscaling. While newer versions of Corona may handle this adjustment automatically, we recommend to verify this setting in your Render Setup to avoid unwanted blurring.
5. Use the ‘Save 50%’ VFB option
Once the render is complete, use the new ‘Save 50%’ button in the VFB to instantly export your final image.
Watch the full video tutorial here
How do downscaled renders compare to traditional rendering?
The downscaling method produces results that are nearly identical to traditional renders, 6x faster.
In the video tutorial, a traditional, high-quality render took 41 minutes to reach a clean, professional state. By using the downscale method, almost the same level of visual quality was achieved in 7 minutes.
When should you use the downscaling workflow in Corona?
You should use the downscaling workflow in Corona whenever you need to produce high-quality architectural visualizations on a tight deadline, though this method is less suitable for large-format print work.
Best use cases for downscaling:
-
Tight deadlines: This method is a lifesaver when you need to deliver HD images in a short amount of time.
-
Mobile & tablet client presentations: Smaller screens — like laptops, tablets, and smartphones — don’t require the extreme pixel accuracy of a billboard.
-
Creative freedom: You can test multiple lighting setups and material styles in the same amount of time it would normally take to create just one image.
What are some of the trade-offs of the downscaling method?
While the downscaling method significantly cuts render times, it can lead to minor losses in reflection accuracy and is not recommended for large-scale printing.
Trade-offs of the downscaling method:
-
Reflection accuracy: Because the Intel AI Denoiser "guesses" how to smooth an image, it can occasionally smudge very fine details in complex reflections on surfaces like glass or polished chrome.
- Large-format screens & prints: While this technique is ideal for smaller screens, stick to the traditional route for large-format screens and prints when every pixel must be physically perfect.
- Animation Flickering: This workflow is currently recommended for still images only. Because the Intel AI Denoiser lacks "temporal consistency," each animation frame may end up with a slightly different denoising pattern. When played back, this can result in visible flickering that is difficult to fix in post-production.
Frequently asked questions
When should you use downscaling versus distributed rendering in Chaos Corona?
You should use the downscaling method when you need a quick result without setting up a render farm. Downscaling is a technique used to speed up the rendering process on a single machine, whereas distributed rendering utilizes multiple computers on a local network to render one image.
Is this workflow suitable for professional client deliveries?
Yes, this workflow is suitable for professional client deliveries, especially for client presentations and imagery viewed on a screen. Some cases — for example, large physical prints or billboards — have the traditional rendering method as the gold standard.
Does this method work for interior and exterior scenes?
Yes, this method works well for both interior and exterior scenes! Since interior scenes often suffer from more noise in shadowed areas, they may see the most dramatic visual benefit from the combination of the Intel AI Denoiser and downscaling.
Does this reduce GPU or CPU load significantly?
The downscaling method does not reduce GPU or CPU load; however, it significantly reduces the total working time of your machine. By finishing a render in 7 minutes instead of 41 minutes, you free up your hardware for other tasks much faster.
How does this affect post-production workflows?
Downscaling simplifies the post-production workflow. With the image being already “cleaned” and resized, you can jump straight into color grading or compositing without extra resizing steps.
Can this method be replicated in V-Ray?
Yes, this method can also be effectively replicated in V-Ray by applying similar high-resolution rendering and downscaling principles.