Just like DaVinci Resolve’s Speed Warp (DR for short) and Topaz’ Video AI (TVAI), Twixtor is now using neural networks aka machine learning to generate additional frames for slow motion (slo-mo) in post-production. We have already reviewed the early beta of its version 8 last year (the link includes examples of the results). But now the software is not only final, it is also available as a standalone version for Windows and MacOS. Any big changes?

Features of the standalone Version

The standalone version of Twixtor is easier to use for beginners than the plug-in, you don’t need any workarounds to extend the clips for slo-mo (see my other article). You can choose a faster or a more precise algorithm and the mode of control between a simple speed change or speed ramping. The latter offers quite intuitive controls you may wish to have in other software, but setting keyframes was not yet completely bug-free. Choices for the treatment of audio are self-explanatory, as is the speed slider. You can define the range of frames to be treated and choose the frame rate for the output independently of the input rate. Slightly confusing is the activation of the GPU (which is normally the right choice), since clicking on the field makes it dark. Now, is that on or off? BTW, the installer is looking very “Windows” on a Mac, but that works flawlessly.

For input, all flavours of AVC, HEVC and ProRes are accepted, but not DNxHR or Cineform, which are also high quality intermediate codecs. They are needed by users of DR under Windows, who can’t export ProRes. Neither is the MXF wrapper accepted, which is an important alternative to MOV. Unfortunately, you don’t even get an explanation when trying to use one of these, the picture just stays black. In the end, the plug-in might be the better choice for professional users, and for log, HDR, or RAW you should also resort to that. But then, RE:Vision FX gives you a license for the standalone together with the plug-in version.

The most serious limitation is in the output codecs you can choose from: there are only HEVC and H264, and both come in 8 bit with 4:2:0 chroma subsampling only. That happens without being mentioned anywhere, you’ll need a tool like MediaInfo (free) to find this out. Not a perfect choice if you want to treat the results further in any NLE with colour grading or filtering. At least you can up the bitrate for that purpose. The fact that they are wrapped as .m4v doesn’t make things better, some NLEs only accepted those after a re-wrap into .mov. And then, they contain no timecode, which can wreak havoc if you have to move files after editing or if you need to send them to collaborators. Definitely add TC while you are re-wrapping for such situations (with a tool like e.g. QTchange).

Setup for the Test

Our earlier tests suggested that higher image quality, more fps and shorter exposure times yield better results with any of these softwares. We didn’t want to make it too easy for the candidates, so this time we have used some smartphone footage. It was shot in HD only, at 30 fps and under available light. Of course, we have chosen a scene with intense activity, in this case dancers and drummers with criss-crossing movements. Get the source file here for your own tests. This is the stuff where trying to get good slo-mo in post usually fails. But, after all, you can shoot good slo-mo in most modern cameras if you have thought of it beforehand.

You can help Twixtor Pro as a plug-in to get pretty perfect results with masks and tracking points, but it’s not really easy to use in DaVinci Resolve or other hosts, as explained in my older article. The standalone version is rather simplified and straightforward, which can be very helpful for beginners. So, we also used standard presets for Speed Warp Better and the model offered in TVAI automatically for slo-mo without scaling, which is now Apollo. Fine tuning can result in even better results from both of these. TVAI is offering only 4x, 6x or 8x of slo-mo, so we decided to try 6x, since 8x seems a bit too adventurous for such footage. We ran the tests on a M1 Pro MacBook with 32 GB under Sonoma 14.7.2.

Test Results

Our test results at the highest quality setting were close, you need to do watch carefully to see the differences. All three are struggling with small objects like the drumsticks, which flicker on and off, and all three can show some pushing and pulling of the background or of criss-cross movement, in particular where there is motion blur. For this less than optimal source footage, the results of all three are still impressive. But personally, I’d say TVAI gets the crown here, in particular for the least background distortion. The faster mode in Twixtor reverts to optical flow, which would also be an option in DR. Both are considerably faster than the AI and can suffice for less critical footage. TVAI also offers a faster alternative for the Apollo model.

Of course, speed also matters, and TVAI’s Apollo model needed only 2:19 if nothing else was activated, the time is predicted quite well (the result is here). So, it needs only 3.5 times the playback time of the final sample to render. This applies once the model is downloaded for the first time, which can need quite a while depending on your internet connection and their servers. Its only shortcoming: you can’t do speed ramping, it works with fixed ratios. But it can generate its output in high quality codecs. Blackmagic’s Speed Warp Better is considerably slower at 7:11, a factor of 11, but the visual results are close (see here). Being part of DR, it handles speed ramps, albeit a bit clunky, and a broad choice of codecs. Twixtor was the slowest at 12:15, which is a factor of 18.8 (the result). All of them are predicting the time needed pretty well, and all three got the GPU cores of our humble laptop fully loaded when rendering, so you can expect considerably better speeds with higher numbers of cores on a Mac or with any strong GPU on a PC.

We tested some very high quality footage in UHD at 50 fps too, you can see a frame of those Bougainvilleas in strong wind in the header above. But the results of the best algorithm in either candidate were so close that only speed would matter. Considering quality and speed, TVAI’s Apollo model was the best, their optimisation for speed in particular is truly impressive, when compared to early versions. But it’s also quite costly at 300 US$ initially, being subscription based. The deal is fair: You are only excluded from further updates if you don’t pay any more, but it’ll still work. And then, it can also do excellent de-interlacing and upscaling (see my article here). Speed Warp is a Studio feature in DR, which costs just the same at 300 US$, but also includes dozens of other valuable features, and until today there was never a charge for updates. Like in our earlier test, Twixtor is still the slowest, even after being out of beta. But at the current price just short of 100 US$ for a permanent license, it is also the cheapest, while you pay as much for TVAI every year.

Recommendation

Twixtor standalone is easy to operate and the best bid for synthetic slo-mo if you don’t own DaVinci Resolve Studio anyway. But it has its shortcomings and is the slowest of the bunch. If you own DR Studio, its internal Speed Warp will suffice for the occasional slo-mo, even if it’s not that much faster. But TVAI is winning the crown here, and not only for quality and speed. If you are planning for slo-mo, you can shoot higher speed in most cameras at a lower resolution, and then let TVAI do the additional slo-mo plus upscaling. Even if this approach takes more time than slo-mo alone, it’ll still be faster than the others and can yield excellent results. But it comes at a hefty price tag.

The post Out on its own: Twixtor Standalone first appeared on DIGITAL PRODUCTION and was written by Uli Plank.

Revision Effects has doubled down on retiming innovation with a major update. Now, there’s Twixtor Standalone, a host-free version of its renowned retiming tool. Second, they’ve included Twixtor ML, a machine learning-powered retiming solution designed to handle motion interpolation like never before.

---

Twixtor Standalone: No Plugins Required

Twixtor Standalone frees users from the constraints of host applications. Its intuitive interface enables editors, compositors, and motion designers to import footage or image sequences, apply Twixtor’s optical flow-based retiming or the Precise mode with ML-support, and export results—all within the standalone application. Twixtor ML (Machine Learning) is where Revision Effects takes retiming into next-gen territory. Using AI to enhance motion interpolation, it delivers higher accuracy for footage with tricky elements, like overlapping motion or fine details. For more details, visit the official page here.

Features include:

• Wide Format Support: Compatible with ProRes, H.264, H.265, DNxHD, and image sequences (PNG, JPEG, OpenEXR).
• High-Quality Motion Interpolation: Maintains Twixtor’s hallmark frame-by-frame smoothness and artifact reduction.
• Speed Ramping and Frame-Rate Conversion: Effortlessly handle varied frame rates or create dynamic speed effects.

---

Pricing:Twixtor Standalone costs 150 Dollars, and has an introduction price of 99 Dollars.

---

The Verdict: What’s Right for Your Workflow?

With Twixtor Standalone, Revision Effects provides a streamlined, independent retiming tool tailored for artists who prefer standalone solutions. Before diving into a new retiming workflow, always test software on project-specific footage to ensure the best results. While these tools promise advanced capabilities, real-world scenarios can introduce unique challenges—especially with highly dynamic shots or unconventional footage formats. For more information, visit the official Revision Effects website here.

The post Twixtor Goes Solo first appeared on DIGITAL PRODUCTION and was written by Bela Beier.

All established methods are based on recognising details in successive individual images and use these to calculate motion vectors. This is then used to calculate the displacements of pixel groups for the intermediate images. This works best if the images have little blurring and good contrast. In addition, such slow motion is usually better if the source material has already been recorded at 50 fps or more. The algorithms primarily have difficulties with uniform patterns where the direction of movement can be incorrectly recognised. They also have problems with segmentation, i.e. distinguishing between moving foreground elements and the background or intersecting movements.

The typical artefacts are “ghost images”, i.e. additional elements where there were none in the image, and the dragging of the still background or opposing movements. We therefore used a hyperlapse over a rice field with a drone as “evil” test material, as in previous tests in the DP, i.e. lots of sharpness but repetitive structures. Then women stamping rice and dragon dancers on Chinese New Year because of the fast, criss-crossing movements. Finally, as a technical object, a helicopter extinguishing a fire, fast-flowing water and a close-up of rice stamping. For the last subject, we had a quadruple slow-motion shot from the camera with reduced resolution.

The opponents

We have already tested Topaz Video AI (TVAI for short) and the equally CPU-intensive “Speed Warp” from DaVinci Resolve (DR for short) (DP 23:01). However, version 4 of TVAI is now available with the new AI model “Aion”, which is supposed to be optimised for precisely this task. On a Mac, it only runs from Ventura onwards and should at least be fed with HD. A new addition is Twixtor version 8 as a public beta, which has long been established as a plug-in for Optical Flow. The new version for the first time also uses a neural network called “DNN – model 1” . As a plug-in, Twixtor has a specific problem: you have to create space for the extended result because DR does not support this as elegantly as its own Speed Warp.

To do this, you can either repeat the clip on the timeline until it corresponds to the new length, or place a coloured area (solid) of the desired length on the lower track and the clip above it. The whole thing is selected and a Fusion clip is created from it. Then add Twixtor and set the desired slow motion in the inspector. Speedramping with keyframes is also possible, and the Pro version also allows masks for better segmentation.

Performance

We tested material in HD and UHD on a MacBook M1 Pro. An eightfold slow motion was calculated, so the software had to create seven synthetic images per real image. The AI processes required significantly longer computing times than conventional Optical Flow. All of them rely on fully utilised GPU cores, the CPUs have hardly anything to do. The values always refer to the result at 40 seconds: In HD at 25 fps, Optical Flow Enhanced Better in DR needs just under a third, Speed Warp on the other hand needs eight times the runtime, TVAI Aion needs just under 10 and Twixtor a factor of 17. With a source in UHD at 50 fps, Aion needs a factor of 74, the relative values of the other methods are similar.

Quality

The examples, some of which are also available for download(is.gd/slow motion files, with the test files and the images), clearly show how much the results depend on the subject. The women stamping the rice are an extreme example: fast, intersecting movements with motion blur and small, complex patterns of clothing remain a challenge for all methods.

Here you have to perform pixel peeping frame by frame to identify differences. Speed Warp, which can achieve amazing results with other subjects, is only slightly superior to the much faster, non-neural algorithm here. Twixtor’s DNN delivers better results, but at the cost of enormous computing times. In our opinion, Aion also looks better than Speed Warp and is only slightly slower.

Our other test series with dragon dancers at the Chinese New Year brought even better results with the AI-based processes. You have to look closely to even notice the deformations in crossing movements or in the background. Here too, Aion looks at least as good as Twixtor. However, it has the disadvantage that this does not yet work in the plug-in for DR, but only in the standalone version. Twixtor and Aion did not quite pass the endurance test with the rice paddy either, with Twixtor showing fewer artefacts with local blurring, while Aion produces large-scale distortions in the same place.

Finally, the helicopter did not show the ghostly rotor blades as with conventional Optical Flow, but instead the rotors became pulsatingly shorter and longer. The AI just doesn’t understand anything about rotors in lateral perspective ;-) However, it also became apparent here that fast-flowing or falling water is rather uncritical. A river shot with intensive movements confirmed this: here, the human eye can hardly notice the weaknesses of the artificial slow motion. A close-up shot in 720p at 100 fps, which we scaled to HD with TVAI and slowed down to twice the length, also looked quite good: The rice flour dusts quite convincingly.

A higher frame rate during recording is therefore the better alternative despite the lower resolution. TVAI can even do this in one go, as it also scales very well. DR Studio delivers similarly good results with SuperScale Enhanced and Speed Warp, but both processes require enormous computing times. Only a powerful PC with a strong power supply and the Nvidia 4090 would help. However, TVAI does not yet use TensorRT in Aion and obviously processes the AI models one after the other, as it is almost twice as slow as DR when doing scaling plus slow motion.

Since differences can only be recognised with extreme pixel peeping, DR is sufficient here if you don’t want to go from “small” HD (720p) to UHD. You should also get good results if the camera is only capable of HD in slow motion and is scaled to UHD in the same way. After all, it is a typical use case that the artificial slow motion is deliberately added in order to limit extreme resolution losses or lack of light in camera based slow motion.

Commentary

In the ratio of computing time to performance, TVAI with Aion beats the new AI in Twixtor. To be fair, it has to be said that Twixtor 8 is still a beta version. In terms of quality, both outperform Speed Warp in subtleties, but the differences recognisable by the general audience depend heavily on the subject. No AI can currently replace real slow motion from the camera, but the combination of high-quality upscaling of the camera shot and a milder slow motion in post looks very good. And here again the link to the material: is.gd/slow-motionfiles

The post The discovery of slowness first appeared on DIGITAL PRODUCTION and was written by Uli Plank.