Apertus Axiom Beta

[mothaibaphoto]

Since it just some days left to deadline for 1k. Euro payment, I want to ask some advice: does it worth for me to go further?
Starting points:
a) I can't afford just a toy. As independent self-investing producer I have no budget for experimenting.
b) I need a real step up over MKIII i have: 4k, Slow Mo, Autofocus. Yes, and all these are pretty useless without RAW, as a lot of cameras already in the market with such specs and I don't want to loose that quality advances.
So, what is the chances

[Frank7D]

I think the following is correct, but anyone correct me if I'm wrong:

The deadline is to lock in a particular price (hopefully a low one thanks to volume discount) for one stage of the production of the camera.
Even if you don't make the payment, if you are a voucher holder you will still be able to buy the camera "at cost." It's just that "at cost" might be higher for you at a later date with no high-volume price break on components.

So you buy your ticket and you take your chances, as the saying goes.

[Sebastian]

The deadline is to lock in a particular price (hopefully a low one thanks to volume discount) for one stage of the production of the camera.
Even if you don't make the payment, if you are a voucher holder you will still be able to buy the camera "at cost." It's just that "at cost" might be higher for you at a later date with no high-volume price break on components.
So you buy your ticket and you take your chances, as the saying goes.

That is correct!

With Phase 1 we are trying to collect enough funds for securing good volume deals for components like the image sensor which in the end makes the AXIOM Beta cheaper for everyone participating in Phase 1.

If you do not participate in Phase 1 now you will still be able to get a camera at-cost later on. But with a worse volume deal so your camera will be more expensive. How much it will cost exactly though is hard to predict as it will depend on when we will order more components and what the volume then will be. Image sensor single unit is around 1300€ without taxes, in the volume deal we get it around 40% cheaper.

So in summary you will never again be able to get an AXIOM Beta for a price as low as now

[kontrakatze]

But in the end, it will be a beta not a production camera. Nothing that I can work with in the field. That's why although I still have a voucher, it is a no go for me. But I think the misunderstanding was on my side, so I back out without any bad feelings.

[g3gg0]

Magic Lantern itself is also "always beta" and people can manage to use it on a daily basis

[a1ex]

FYI, I'll be meeting the Apertus team in Vienna in about 2 weeks; they promised me one of the first early betas.

I'm not going to promise it will be better than a 5D3 from the first iteration. But in 1-2 years... who knows

[AndreyM]

Probably same as mothaibaphoto I participated crowdfunding and supported Axiom with 350 €.
While the idea is so great, implementation is little slow so I decided not to move forward with phase 1 as of yet.


Will see how it is progresses. Meanwhile I switched from my lovely 5DIII to Sony A7sII which has greal low light, 4K p30 internal and FullHD p120 but of course I'm missing RAW.


Best Regards,
Andrey

[mothaibaphoto]

Great, A1ex finally got it !!! Congrats! This camera is very close to, but not yet what I need: compact, with ultimate possibilities, and without any compromise on image quality, able to shoot, lets say, 500 frames in burst and I agree to wait while it dumps. It's very interesting, what features will finally get this camera.

[Kharak]

Can't believe I left Vienna today!!!! And the Master himself, A1ex, is there!

Had I known.. Could have stayed for another 2 days if I wanted

Would love to meet A1ex and shake his hand, I would never wash it again

[jtvision]

Hi Alex,
Were you  able to look at FPN on blackmagic 4k files I submitted over a month ago on this post?

[Audionut]

This post?

[jtvision]

Yes, last one on that page. I posted noisy Blackmagic 4k dngs hoping for a fix.

[jtvision]

Hi Alex,

Did you have any chance to look at my 4k BMPC dngs?

thanks,

[Kharak]

Yes, last one on that page. I posted noisy Blackmagic 4k dngs hoping for a fix.

.... Are you blind?

[jtvision]

Sorry, don't know what you mean. I dont see any tools for correcting bmpc4k dngs on that page. Or am I missing smth?

[jtvision]

This was tweeted asking if anyone interested to have a tool to fix FPNs. And now being ignored. 

[a1ex]

Sorry, did I promise any release date for the tool?

on-topic: I've looked at the files a while ago, but the existing code didn't work on them out of the box, so it may take a while.

[jtvision]

Cool, Alex! Thank you for taking time to look at bmpc 4k dngs. No worries and no pressure! Take your time. Its just comments of others made me think that I was missing something. 

FYI: Black magic cameras have preferred temperatures to work. FPN appears only if camera is not warmed up enough or if it got too hot after long hours of being on.

[a1ex]

That's a good hint. Are your sample files taken at temperatures outside the preferred range? Didn't find anything relevant in exif.

If so, I think you should be able to fix them by averaging a few seconds of dark frames (a clip with lens cap on) at roughly the same temperature as the clips. Then, use that average as a dark frame in your raw processing software.

For the sensor used by Apertus, my plan is to have a set of calibration files that also covers extreme temperatures, but I didn't put the camera in the freezer and oven yet

[jtvision]

Yes they were shot outside prefered range, after about an hour of continuous shooting.
Thanks for the hint I'll try that. Any noise correction tool you would recommend?

[a1ex]

Here's a cool trick I've tried to implement a while ago on 5D3, without success, but was very easy to do on the Axiom Beta: compressed 4K RAW video over HDMI, with... an external 1080p recorder that doesn't support raw video natively.

This particular implementation maps 12-bit 4K UHD RAW (3840x2160) at 30 fps to 8-bit 422 1080p at 60 fps, on Atomos Shogun.

http://www.apertus.org/axiom-beta-uhd-raw-mode-explained-article-may-2016



The math tricks from there can be reused on the ML side as well, to compress the 14-bit raw data to 10-bit with minimal loss in image quality (nearly lossless at higher ISOs)... if we manage to understand how to program the DIGIC image processor to apply a curve to the raw data.

[Levas]

Brilliant!
Recording UHD with a 1080p60fps recorder

Bayer data to rgb with double frame rate, altering between green channels and pixelshifting red and blue each frame. When I think about it, don't you mis some green channel resolution?

[a1ex]

The two tricks you mentioned can recover the resolution in all channels - see for yourself:

https://www.apertus.org/sites/default/files/reschart-frame00001-original-raw12.jpg
https://www.apertus.org/sites/default/files/reschart-frame00001-recovered-uhd.jpg

How it works: https://wiki.apertus.org/index.php/Reversing_digital_filters

Side note: I've just applied the algorithm I've used to recover the raw data from the HDMI clip on the telephone test images from http://petapixel.com/2013/09/29/newly-developed-software-helps-get-high-quality-photos-crappy-lenses/ or http://www.cs.ubc.ca/labs/imager/tr/2013/SimpleLensImaging/#results

The result surprised me as well:



(their original, their recovered image, and my recovered image from their jpeg)

[DeafEyeJedi]

Wow, that's pretty darn good actually!

[Levas]

ah, there is some more stuff going on, you need to reverse digital filters.
And still, the results are ridiciously good!

(And I learned that it is possible to reverse a digital blur filter  )