Apertus goes 4K, KinoRaw to provide workflow
Recently Apertus, an open camera project, announced plans to create a new camera built on top of a 4K sensor.
The news has already seen a splash of interest in relevant forums, not in the last place because the project is clearly very ambitious. Perhaps too ambitious? Let’s try to get to the bottom of it.
There’s not what one might call a plethora of 4K solutions, and open hardware would certainly be a novel thing. But Apertus is not on their tod with this project. The team, who already worked with Elphel Inc. before and built the current camera on top of their design, has partnered with Dynamic Perception whose motion control systems would be supported by the new camera.
Together they are trying to design the world’s only open high speed Super35 global shutter 4K CMOS sensor camera with an estimated price of less than $10K. A discussion is currently taking place about a possibility to establish a USA based Apertus Foundation to handle funds as the plan is to go for crowdfunding.
From software side the team is now backed by another crowdfunding project, KinoRaw, that aims to provide an end-to-end Blender based workflow solution for the Elphel-based camera.
We asked the Apertus project several questions and additionally contacted Andrey Filippov of Elphel Inc. to find out how things look from his perspective.
On the Apertus end: Sebastian Pichelhofer
A year ago Red published a short video of Ted Schilowitz explaining “Why high Definition is the Low Definition of tomorrow" and why 4K is becoming a must. What features of the future 4K sensor based Apertus could make it look compelling for those diving into the digital and comparing available options such as Red Epic or upcoming Canon EOS C500 and 1D?
If you look at recent press releases from big companies you might notice the lack of real information or technical details. Marketing departments are often able to spin slightly re-worked features as something entirely brand new by making up new words, backed up with doctored images and charts which focus only on the most positive changes that seem to create the impression of great leaps in development.
For industry professionals this is a very frustrating development as they need to invest a lot of time to find out what the camera actually does by reviewing the device from each manufacturer in person or relying on trusted reviewers. Some of these reviewers are approached by the big manufacturers to create demo footage or entire films to promote their gear for them...
This is exactly why I fight for open hardware and free software: honesty. We are not afraid to explain what exactly happens inside our camera, after all Apertus is also about open knowledge and open education.
So what we are doing now is trying to find a good balance in the material for our crowdfunding campaign between simple visual feature overviews with easy to understand description and more in-depth texts for those who want to learn more or are interested in the technical details.
What technical challenges do you envision with regards to the 4K sensor?
Well, the biggest challenge I envision is the incredible amount of data it will produce and that we need to process in real-time. 12 Megapixels in 12 bit mode at "just" 100FPS (the sensor can do more) result in 1,800 Megapixels per second throughput. The current Elphel camera can process around 80 Megapixels/s.
How are you going to solve this?
We will release the full plan soon, stay tuned ;)
What about certain global shutter limitations?
We will evaluate everything once we have an evaluation board from CMOSIS. Just from looking at the sensor datasheet noise might be a problem, on the other hand the dynamic range and light sensitivity is really not bad, even when compared to ERS sensors, so noise might not be that much of a problem after all... We will see.
And the actual designing?
Sure, it will be a big task and challenge, but looking back to the last couple of months (we started with the idea for the new camera in Summer 2011) and how the Apertus members have collaborated and progressed over time I am very confident we will create something amazing. We have a great team with incredible talent.
Earlier this year you published a concept of your own camera control app, Open Cine. What's the exact status of the project?
Unfortunately Open Cine is still a concept, so far we were not able to assemble a team of volunteers to work on it, but we hope that will change once we publish further plans of our new camera or ultimately we will try to include its development in the crowdfunding plans.
What are the overall project's priorities?
Currently the main priority is finishing the material that will be using in our kickstarter campaign. We want to prepare the campaign properly and in a very complete way. If it still fails, we can then go down with dignity ;)
The KinoRaw project
OK, KinoRaw folks are not the most talkative people, but fortunately they wrote a more or less detailed project description at the fundraiser page that provides a generic insight what the project is about.
What the KinoRaw project hopes to deliver is:
- Extensive testing of the Apertus camera
- A research of linking Blender with Elphel for easy editing
- Code development based on the above premise
- Technical manuals for use
- Filming footage exploiting its full potential
- Editing and postproduction of filmed contents
- Presentation and dissemination of results
The project statement cries for elaboration on technical details, but there’s very little of it. According to the googletranslated blog, the team is currently at the R&D stage. There is a mindmap in Spanish available too. We contacted the team and were told that there are also some scripts for Blender in the works, but no public repository yet.
However the fundraiser is currently at 1120 euro of the minimum 3800 euro goal, and there are mere 6 days left. Given the dynamics, the project is very close to failing. Which is, of course, entirely up to you as the community.
The KinoRaw team also hasn’t made any public statements regarding the 4K initiative yet, but as the initiative looks like a midterm goal, the result of their work is likely to continue being useful for quite a while.
We asked Sebastian about that, and he replied:
The KinoRaw Blender workflow (and their current crowdfunding campaign) purely targets the Elphel camera, there is so far no plan for a collaboration on the new planned 4K camera, though we would be open for discussion once we published our entire proposal.
Apparently there is some syncing of plans to be done between the two projects.
Elphel’s perspective, Andrey Filippov
Elphel, Inc. is a USA based company started in 2001 to provide high performance cameras based on free software and hardware designs. Both software and hardware designs are available under GNU GPL. The company's projects today are Elphel 353, a 5Mpx CMOS reconfigurable network camera, and Elphel Eyesis, a panoramic camera system.
What’s your view on Apertus and the idea to build a digital film camera on top of it?
We are interested in this project, that’s why we helped them with cameras in the past. And we are really glad that they are self-supporting now, because we don’t have a bottomless supply of hardware. Elphel is a really small business — we are just three persons who do a lot of expensive hardware development, but sales are our only source of income.
How much do Elphel and Apertus projects interact these days, if at all?
Apertus is more like a traditional free software project, while we are a small business. We do interact — after all, they mostly got their cameras from us. We actually implemented some of their requests regarding the firmware, but, again, we have limited resources.
Do you think KinoRaw is on the right path? Would Blender really be the best key component of the proposed workflow?
I’m not a professional in this field, but I do believe that they‘ve made the right choice. As for Elphel, we do all the post-processing with ImageJ for which we wrote a plug-in.
Elphel’s camera has many real-life applications. What project currently has the top priority for your company?
Right now we focus on the big camera that captures data with 26 sub-camera modules for subsequent 3D reconstruction. We’re having some rather nice progress: image processing is quite OK, so are lens distortions correction and precise calibration of the whole camera (root square error of pixels-to-lines mapping is ca. 0.1px). We will be presenting this camera at SIGGRAPH this year.
Was it useful? There's more:
- Interview with Thomas Krijnen on IfcOpenShell, Blender and architecture
- Interview with Sintel's art director David Revoy
- 3Delight exporter for Blender updated to feature new global illumination method
- Customized MyPaint gains animation for a new open movie project
- Recast & Detour project getting ready for Blender 2.60