Autopano Video - FAQ - Autopano Video 3
- 1 Intro
- 2 License
- 2.1 How can I try Autopano Video Pro 3 Alpha?
- 2.2 What version of Autopano Giga do I need to work with Autopano Video Pro 3.0?
- 2.3 What if I buy an Autopano Video Pro license today?
- 2.4 What happens with Autopano Video Standard (AV - Non Pro )?
- 2.5 What is the price to upgrade to Autopano Video Pro 3?
- 2.6 I own a GoPro Omni rig, how can I get Autopano Video Pro 3.0?
- 2.7 What if I buy a GoPro Omni today?
- 3 D.WARP
This is a FAQ for Autopano Video 3 and it's new features.
New questions will surely be added so be sure to check back regularly!
How can I try Autopano Video Pro 3 Alpha?
If you are a registered Autopano Video Pro 2.x user you can use your AVP 2.x license key available in your customer accountto register AVP 3 Alpha - and all further pre-release versions that will be available soon. Note that the registration is temporary: when the final version will be released, the Alpha/Beta/RC versions will stop working.
If you don't have an Autopano Video Pro 2.x license key then you can try AVP 3.x thanks to the trial version that comes with a few limitations.
What version of Autopano Giga do I need to work with Autopano Video Pro 3.0?
What if I buy an Autopano Video Pro license today?
Starting today, May 17, 2017, all Autopano Video Pro licenses purchased from our store or our authorized resellers will give access to every version of Autopano Video Pro 2 and Autopano Video Pro 3.
Your Autopano Video Pro 3 license key will be emailed to you in a couple of weeks, you do not need to claim it.
What happens with Autopano Video Standard (AV - Non Pro )?
Stitching videos requires a lot of processing power. To deliver an enjoyable user experience and satisfactory rendering time, Autopano Video needs to use the processing power of the graphics processing unit (GPU). Autopano Video 1 and 2 were only using the central processing unit (CPU) which leads to suboptimal and sometimes frustrating user experience and performances.
We thus decided to stop selling a CPU only version of Autopano Video: as of today it is not possible to purchase Autopano Video 2 licenses anymore.
What is the price to upgrade to Autopano Video Pro 3?
We took the decision to offer a unique upgrade price whatever the version you currently own. To upgrade to Autopano Video Pro 3.0 from either Autopano Video Standard or Autopano Video Pro, 1 or 2, the unique price is 269,00 € before taxes.
You will be able to upgrade from your customer account in a couple of weeks.
I own a GoPro Omni rig, how can I get Autopano Video Pro 3.0?
Your GoPro Omni rig has been delivered with an Autopano Video Pro 2.x license, you can thus register AVP 3.0 Alpha and use it without limitations or watermark for a while. You will then have to upgrade to use the final AVP 3.0 version. You will pay the same price as customers who purchased the license separately – see previous paragraph.
What if I buy a GoPro Omni today?
If you buy an Omni on or after May 17 your Omni will be delivered with an Autopano Video Pro 2 license and a free upgrade to Autopano Video Pro 3 when it is released.
What is D.WARP?
D.WARP is Kolor's answer to stitching issues. It is a parallax compensation algorithm. This means that computer vision automatically fixes the viewpoint differences between two videos. It goes beyond just fixing parallax, it also helps with synchronization issues and rolling shutter glitches. We believe D.WARP will also allow to fix most of the stitching issues coming from your shots and this will dramatically reduce the time needed to fix stitching issues during post-production.
Is D.WARP a motion flow stitching algorithm?
Not really. Motion flow based algorithms are computing a complete motion flow on the images, which is costly and can fail when there is no solution (something visible in one camera can be masked in the other one).
D.WARP is smarter than that. It only examines the overlapping area between cameras and computes a deformation in source image so that both images fit on the seam.
This approach has 2 advantages, the first one is that the deformation (or warp) of sources images is smoothly diffused toward the center of images, avoiding too much pixel stretching like what is sometimes visible on some motion flow implementation. The second is that the approach is time-coherent which makes the algorithm relevant to moving cameras or subjects. Simple motion / optical flow approaches fail in both areas.
Will it fix any parallax?
The amount of parallax it can fix (closest object distance) highly depends on the size of the overlapping area. In shots made with Omni, we can correct objects that are at least 60cm away. Again, we expect this algorithm to work on many cases but not all cases. All feedback on failure cases will help us improve D.WARP.
Does it work with moving cameras?
Yep, D.WARP works for both static and moving shots. Later we will add some settings to tune the D.WARP algorithm that will enable you to get even better results for all situations.
What type of new artifact will D.WARP create?
As D.WARP is mainly a warp mesh on the source videos, you can see artifacts that you can expect from such approaches e.g. backgrounds "bumping" around fast moving objects entering the frontiers or bent straight lines.
Note: In many cases, D.WARP won't deteriorate the content quality of the source the way motion flow algorithms do.
Is D.WARP compatible with any rig?
Currently D.WARP only works with the OMNI rig. The reasons are the following:
- The geometry is known
- The synchronization is perfect
- The overlaps are big enough for D.WARP
Omni was the first choice for D.WARP, but our aim is to make D.WARP compatible with more rigs. Stay tunned for updates on that.
Is D.WARP compatible with Autopano Giga?
The D.WARP feature cannot be used in Autopano Giga yet. It means that you can still edit your stitching in Autopano Giga but you won’t be able to preview the effects of D.WARP. Make sure to delete the markers from Autopano Giga as they are not compatible with D.WARP. Using the D.WARP feature should be the last step in your workflow.