E-Cycles

On Sale! by Mathieu in Scripts and Addons

Focus on your Art

New High-Quality Pass mode available in E-Cycles 2.91 and 2.92!


E-Cycles 2021 alpha builds: the first new feature is already available. You can use SSAA in the viewport for very sharp rendering of fine details and crisp textures. It also offers slightly faster rendering at same sample count! The first stable build of E-cycles 2021 will be released in about 3 months, when 2.92 is ready.

Comparison of the same scene with and without SSAA after 10 seconds of viewport rendering using OptiX denoising. Made using an RTX 3090 and E-Cycles 2.92 alpha.

E-Cycles is the fastest photorealistic renderer for Blender, offering up to 2.5x faster path tracing out of the box. Installing it is as easy as unpacking a zip, and it’s 100% compatible with Blender/Cycles. Get it now to save money and a huge amount of time. Using the fast profile, the RTX version is up to 41x faster than even the OptiX version of Cycles!

"E-Cycles lets me render & proof huge print ready assets locally on 3 gpus – no renderfarm rt required!" – Paul, CG Artist

  2x faster rendering out-of-the-box, 13x faster with same look, 41x faster for fast previews. All variants made with a maximum of 3 clicks. Scene is made by Khánh Nguyễn Nông Ngọc.  


User tests done by Hyesung in October 2019

E-Cycles is available based on Blender 2.79 to 2.92, with updates until the end of 2021. 

Render by Petra Mikuličić, VDL Pavilion presented by Kettal, designed by Dion & Richard Neutra
 

E-Cycles is extremely fast and flexible

Path tracing gives great results, but it’s often slow, has a lot of settings that need tweaking, and requires expensive hardware. Render farms are an option, but they need long uploads and downloads, are limited to certain Blender versions, and are expensive in the long term. Real-time renderers like Eevee are fast, but they have a limited quality level in terms of lighting and complexity. They also require a lot of manual tricks, which tend to break when the scene is changed – sometimes even with a simple camera move. You also need to tweak the scene when you come from Cycles.

E-Cycles is fast, stable and simple to use. Out of the box, most of the time it's like doubling your GPU power. So, if you have an RTX 2070, you get around €400 worth of GPU power; with a 3090, over €1500 worth of GPU power. By selecting a render preset and using the new Persistent Data option, you can render at around 3 seconds per frame on a single GPU.

E-Cycles is super-fast and flexible. In just a few clicks, you can switch from Eevee quality level (although it’s often faster than Eevee) to highly realistic rendering. Perfect to adapt to your client's needs with a single solution. The faster presets are almost real time!


Scene by Entity Designer and rendered with E-Cycles

Which version is best for me?

The standard version of E-Cycles works on any CUDA capable card supported by Blender. It's about 2x faster and offers the full features of Blender CUDA with adaptive sampling and dithered sobol on top. It's the recommended version if you have a GTX card, or if you need features that are only available with CUDA in all your projects.

The RTX version includes the standard version, plus a super-fast version specifically made for RTX cards to take advantage of NVidia's hardware acceleration. It uses OptiX, and as such also has the same limitations (see "E-Cycles RTX known limitations at the bottom). As the standard, fully-featured version of E-Cycles is also included, you’ll have the best tool for every project. This is the recommended version if you have an RTX card.

The Nebulae version is up to 16x faster out of the box, but only works to render Nebulae. This is especially useful if you’ve followed Gleb Alexandrov's space VFX tutorial, or bought the Nebulae Generator. This version will only be updated for stable releases of Blender.

How does it work?

Thanks to code optimizations (for efficient memory access and better usage of the massive core count of GPUs), using E-cycles is on average like doubling your GPU(s) count. You can just hit render and immediately benefit from up to 2.5x faster path tracing with the same image quality.

Some new optional features allow up to 41x faster path tracing. Thanks to a new preset system, you can quickly switch between good looking renders done in seconds and high-quality photorealistic renderings. Does your client want more realism? With just two clicks to select a profile, you can hit render and you’re done. 

The time to render the following scene went from 5 minutes to 9 seconds on a single 2080Ti. A tutorial and the appropriate presets are included, so you can quickly get the same results.

Tibor Nyers from Boostclock.com even got this down to 5 seconds per frame, including preprocessing, rendering, compositing and file saving: https://blenderartists.org/t/e-cycles-faster-rendering-for-cuda-gpus-and-better-ai-denoising/1139717/898

With E-Cycles, you get support and weekly builds based on the latest and greatest Blender version. E-Cycles is available for Windows, Mac and Linux*. Both versions are a perpetual license.

E-Cycles helps you:

- save a lot of time and money by rendering faster with the same quality. The average speed increase for the path tracing phase is around 2x out of the box when using a CUDA GPU. Thanks to a new sampling strategy, the noise level is even lower at the same sample count when compared to vanilla Cycles

- render extremely fast fly-throughs by computing the pre-processing phase only once when doing a fly-through (where only the camera moves)

- improve your render quality while rendering in the same time as before, using (for example) more bounces and caustics

- or go to the next level with videos and VR, thanks to new presets that render up to 15x faster

- render extremely clean preview images, thanks to a new AI denoiser that gives great results on still images with as little as 16 samples. Really helpful for quick feedback with your clients, as well as for baking

Example using E-Cycles AI denoiser and 4x less samples to render 4x faster while retaining more detail. Comparison by Marvin Lübke. 

- save yourself time by using auto tile size and quick quality presets. Shorter setup time and faster rendering means it's easier to adapt to changes and deliver projects on time. E-Cycles can really save the day on tight deadlines

- save a lot of money on your electricity bill, thanks to shorter render times

- get more out of your current GPU and keep your hardware on a budget. Buying another 3090 costs about €1500, for example, and you may also have to buy another power supply (+€100). So in this case, E-Cycles can save you 1500+100-299 = €1301

E-Cycles is easy to use

Render faster out of the box with the same quality:

2.5x faster rendering by just clicking "render" and 4.4x faster after 2 clicks to select a profile. 

Or boost the performance with just 2 more clicks to select a preset – here for example 4x faster:


Impressive AI denoising

Render good-looking previews with the new AI denoiser, even at very low samples (16 samples here; original scene from cgstrive):

The original noisy image:

And here an example of the denoiser used in a video for Archviz:

Here you can see a comparison of the 2 denoisers with fluids and volumetrics from Alleyne Studio:


E-Cycles is regularly updated:

On top of the weekly updates, here is the changelog from January 2019 to now:

February 2019 update

  1. New Dithered sobol for cleaner render at low samples.
  2. up to 5% faster rendering in some scenes.


March 2019 Update:

  1. Cleaner Viewport renders in complex scenes.


April 2019 Update:

  1. Intel Open Image Denoise Patch from Stefan Werner is included.
  2. Latest Filmic log encoding for top-notch post-pro as recommended by Bartek Skorupa https://www.youtube.com/watch?v=kVKnhJN-BrQ.
  3. New option based on the scrambling distance patch from Lukas Stockner, allows to render up to 40% faster (on top of existing speedup).
  4. Uses half the memory when denoising with all passes (transmission and SSS included).
  5. New AI denoiser addon which can now generate denoising node trees in one click, even in complex scenarios and with very high quality.


May 2019 Update:

  1. E-Cycles 2.8x got 2 improvements ported from the 2.79x branch. Memory usage is reduced in most scenes, allowing to fit bigger scenes in the GPU memory. Rendering is also around 10% faster compared to the April update.
  2. Both the 2.8x and 2.79x versions can render up to 2x faster when using the old denoiser (the one from Blender) on multi-GPU setups.
  3. Updated to latest Open Image Denoise, using 38% less memory to denoise.


June 2019 Update:

  1. E-Cycles will automatically choose memory efficient code path depending on the scene, using up to 40% less memory than master/buildbots (BMW scene on a 2080Ti goes from 2GB on buildbots to 1.2GB memory usage in E-Cycles).
  2. An option to force one memory efficient option has also been added. It makes transparent shadows optional. Render times can be greatly reduced in some scenes.


July 2019 Update:

  1. The AI denoiser tree generation has been improved and has a better auto layout now.
  2. Adaptive sampling is now available for CUDA and can bring some good speedup in selected scenarios. Own test on smoke+fire shows up to 2.5x faster rendering compared to the June update. (Although only CUDA is officially supported, OpenCL users also report noticeable speed-up)


August 2019 Update:

  1. Based on the official 2.80 version.
  2. AI denoiser updated to 1.0. It allows to denoise renders of any size with low memory usage.
  3. New buttons for quick background command line render. You can then close the UI, which gives better stability, makes more memory available for your scene and also speed-up render in many case.
  4. Baking render speed was optimised in some scenarios.


September 2019 update:

  1. Completely reworked AI denoiser:

   - faster by default, particularly on multi-GPU setups

   - much better memory management thanks to a patch by Stefan Werner

   - streamlined denoiser UI

   - automatic optimizations of the denoising tree. Improves denoising and rendering speed.

2. Faster baking:

   - Baking is now up to 20x faster by taking smart decision for optimal performance.

   - Baking will require a bit less memory too.

October 2019 update:

  1. Added a remove double modifier.
  2. Added an experimental bridge modifier.
  3. Added bevel and crease weight to the boolean modifier for non-destructive and easy to use bevel after boolean.
  4. Added light groups patch from Lukas Stockner (experimental).


November 2019 update:

  1. out of core Rendering for RTX (speed limited by system Ram, may require Quadros or Teslas for good performance)
  2. much faster rendering for RTX
  3. further optimizations for CUDA


December 2019 update:

  1. Pre-processing optimizations
  2. Update stable build to 2.81a
  3. Adds 2.82 builds


January 2020 update:

  1. Viewport denoising for the RTX version (original patch from P.Mours @Nvidia)
  2. Add experimental support for GTX cards in the RTX builds, including viewport denoising.


February and March 2020 update:

  1. Very fast volume rendering
  2. OpenVDB import preview builds
  3. Update stable build to 2.82a


April 2020 update:

  1. Faster viewport rendering
  2. AO and Clay kernel for very fast previews
  3. NVLink support to stack memory on multi-GPU configurations. Allows to nearly double the size of your scenes while being much faster than with system memory.
  4. New presets with improved performance and quality
  5. Improved UI
  6. easy switch between CUDA and OptiX


Mai 2020 update:

  1. Slightly improved performance
  2. Better Clay Preset


June 2020 update:

  1. Faster Denoising
  2. Updated OIDN for higher quality denoising
  3. New splash screens and logo to easily identify the version you work with.
  4. early access to the new sky model.


July 2020 update:

  1. New tutorials for up to 41x faster rendering, higher quality rendering, general overview of workflows and functionnality, tips and tricks, etc.
  2. Denoising quality when using the slow pass mode was improved.


August 2020 update:

  1. New Light Groups add-on (included for free)


September 2020 update:

  1. Improved Light Groups add-on. You can now easily change the tint of your lights and apply both the strength and ting to the lights themselves for final rendering.
  2. The integration of the Light Groups Add-on with the AI-denoiser was improved.


October 2020 update:

  1. Light Groups can now be tweaked directly in the viewport. Previously tweaking a light would reset samples count to 1 and viewport resolution to 64x64. E-Cycles 2.91 now offers the possibility to create the perfect light setup really quickly thanks to high sample count, full resolution and real time updates!

2. Persistent Data option: this option will export the scene only once per animation. Especially usefull for fly-through and turntable animations of complexe scenes.


Use all your add-ons

E-Cycles is 100% compatible with Blender and is synced every week with the official version. It means you can use all your beloved add-ons for the version you choose (2.79x and/or 2.8x). Here is an example from Alleyne Studio using Graswald and Pro-lighning skies with E-Cycles:


Users of E-Cycles:

Alleyne Studio, you can see several screen recording of him using E-Cycles on his Youtube channel.

Giacomo Soriani http://giacomosorianistudio.com

Mike Pan is also using E-Cycles here https://twitter.com/themikepan/status/1120393119286145027 , getting faster rendering and cleaner images.

A good sign also is that many users are happy switching from Octane to Blender: https://blenderartists.org/t/e-cycles-available-on-the-blendermarket-50-off-this-week/1139717/1219

https://blenderartists.org/t/e-cycles-available-on-the-blendermarket-50-off-this-week/1139717/1242

https://blenderartists.org/t/e-cycles-available-on-the-blendermarket-50-off-this-week/1139717/1243

https://blenderartists.org/t/poll-what-do-you-render-with-2019/1156640

Flavio Della Tommasa, original post here:

Obviously in terms of performance I believe the data speak for themselves. On 8 images, 2h each, I lost a day of calculation. With E-Cycles, in just over 2 hours I solved everything.

Robert Brian about the support, original post here:

I have few thoughts about E-Cycles and now I’d like to share especially those who are unsure about E-Cycles.

Past week I asked Mathieu to include a patch into E-Cycles. He promised to do it after he is finished with the May update. I didn’t have to ask, a PM was waiting today and he wrote it’s already included. And it’s working. He was and he is very supportive all the time. And this was true even when I was not a potential buyer. So I do not regret the purchase of the perpetual license at all. Keep going this way!

Lissanro Rayen, original post here

Impressive speed-ups, support is good, all issues (specific to E-Cycles build) I have found and reported were quickly fixed

JulioGB, original post here and here

Thanks for your fast replies, your support is fantastic. The only thing i dont like about E-cycles is that i haven’t used it earlier

Komposthaufen, original post here

The Speedup is amazing

Wolfie138, original post here

3m:09s to 1m:42s, needless to say, I’m impressed!

Rendered with E-Cycles:

Renders by EntityDesigner:

Renders by Albert Shamseev:


Renders by Helmy Ardiansyah - STUDIOHV:

Rendered in 23seconds per frame on a single 1080Ti:

About the Author:

My name is Mathieu, I work on Cycles since 2015 (my first patch here https://developer.blender.org/D1530) and already have several patches in official Blender.

What about OpenCL and CPU?

Only CUDA GPUs are officially supported. 

For AMD GPUs users, OpenCL is only around 20% faster at the moment and is still being worked on. CPU is like in official Blender. You can still benefit from the new AI denoiser on all platforms. You can see the AI denoiser compared to other ones here and here or a 7sec render of the BMW scene with AI denoising here. Here you can see an AMD user using E-Cycles successfully too.

If you want to use E-Cycles with CPU or OpenCL, I gathered some user tips in the documentation to get the best experience possible.

Pre and post-processing:

E-Cycles on average is like adding another GPU to your configuration. Just like another GPU would, E-Cycles will make the rendering phase faster. Pre-and post-processing are done on CPU in Blender and will be as fast as before in E-Cycles.

Known limitations:

- Support is only for the E-Cycles parts. Bugs found in buildbots are to be reported to the Blender Foundation on developer.blender.org 

- Scenes have to fit in the GPU memory for the speed-up to work properly. Since some months, Blender supports rendering scenes on GPU while using system memory to some extent. While E-Cycles allows this too, the speed of the system memory is too low and your GPU will just wait for datas to arrive. In most cases, scenes requiring more memory than available on the GPU should be rendered on CPU

- CPU+GPU works, but is also not officially supported. There are many cases where big tiles are still required for GPU rendering, while CPU rendering requires small tiles. Only the GPU will be faster with E-Cycles. The documentation lists some user's tips if you still want to use your CPU for rendering with E-Cycles and want optimal performance in this scenario.

- Multi-GPU configurations and denoising: their is a bug in Blender 2.8x currently when using multiple GPUs with CUDA and the old denoiser or it's passes. It can slow down rendering. Please use the AI denoiser instead at default settings if you have multiple GPUs.

- Low samples renders (under 64 samples per pixel): thanks to very high quality denoising, you can get better results/detail preservation with E-Cycles at low sample count. However, because of the scene export times taking a good part of the overall render time at low sample and GPU requiring a lot of work to be efficient, direct speed-up of path-tracing at very low sample count will generally be negligible. At comparable noise/detail level, E-cycles is still around 2x faster, but at low samples you may have to tweak your sample count and denoiser setup manually. Earlier version of E-Cycles had automatic sample count for such case, but it was decided after feedback from users to let the user decide if they want a cleaner render or a faster one in this case.

- The out-of-the-box speed-up is mainly for Path Tracing. Branched Path Tracing may also benefit from a speed-up, but it may vary depending on the scene.

- E-Cycles is shipped with a full version of Blender until there is a proper render Engine API allowing good integration of external renderer.

E-Cycles RTX known limitations:

Those are the same as Blender OptiX, I list those here for convenience.

- Requires a recent Nvidia driver version 436+ (New! GTX cards are now supported)

- Some features are not supported. If you need those, please use regular E-Cycles (included when you buy the RTX version):

  - Volumetrics (works, but may be slow),

  - Branched Path Tracing

  - Raytracing nodes (Bevel and AO)

  - Baking 

- first render takes more time due to kernel compilation.

- no CPU+GPU rendering

The standard version of E-Cylces is included in the RTX variant, so that you can always use the best tool for each of your projects.

System requirements:

Same as Blender plus:

- Any NVidia GPU supported by Blender. (Although only NVidia GPUs are officially supported, many users now use E-Cylces on CPU or AMD GPUs. I added some user tips in the documentation about it)

- a 64bit version of Windows, MacOS** or Linux supported by Blender.

*The Linux version of official Blender builds is faster as the Windows one. For E-Cycles, all systems have about the same speed. So the speed-up may be relatively smaller on Linux by about 10-20%. It means the mean speed-up should be around 1.7x instead of 2x.

** For CPU rendering, any MacOS version supported by Blender will work with E-Cycles too. For GPU rendering, please check with official Blender first to see if your GPU is supported for Cycles rendering (in edit-> preferences -> system -> device). For CUDA rendering MacOS Mojave is required. Blender 2.90 will probably drop CUDA support on Mac.