EPIC and Alexa...

[Hans von Sonntag]

Really, the great shocker here is the ~1.6K resolution of Alexa.

Personally I'm not too much into the resolution is all thing. I fully understand that with the development of digital cinema resolution will increase further but resolution is just a part of many things that make a picture attractive.

We have perceived sharpness, we have DOF, we have colours (especially skintones), last but not least ergonomics, etc... This all factors into the decision of the filmmaker which camera she/he chooses for the project.

I did not buy my RedOne because it can resolve approx. 3.2 K. I bought it because I can use my most liked lenses on a camera that works very well in the 1080p/2K realm I work in and safes me tons of money on film stock that I can invest now into more important things like talent, for instance.

Alexa could do the same for me. It just happens that I'm already on the Red train and won't jump off. Epic is too attractive to miss.

Hans

 

[Sanjin Jukic]

Thanks Jim for this great, valuable and unbiased test.

I do not understand why some of professional cinematographer groups are so loud about Alexa

and also why they have all the time negative opinions towards RED MX!!??

Alexa is HD $70,000.00 and 2K $130,000.00 camera in a competition with Sony F35 and Panavision Genesis.

From what I saw Genesis still have a lead towards Alexa and F35.

RED MX is overall leader just because of its image quality, resolution, easy of use, affordability and accessibility.

There are more RED NX already than Alexa, F35 and Genesis together in this world.

And the number is rapidly growing daily.

Upcoming Epic is going to be a real epic that with its resolution, size, lightweight, easy of use, image quality, modularity,

upgradeability, affordability, accessibility and future proof capability will turn overall industry totally upside down.

Somebody totally have got lost here but for sure it is not RED DC with R1-MX, Epic and Scarlet.

 

[Peter Majtan]

Hans, if You have followed some of my other posts over the months (years now), You would know exactly where I stand when it comes to resolution. My reasoning to see some clarification from ARRI is also not for the purpose of comparing it with the Epic. That (comparing 5K to HD in terms of resolution) is pointless. As this is the first resolution chart I have seen from Alexa, I am simply puzzled by what I see. 3K RAW should downscale to 2K perfectly and I know ARRI's sensors - this is not at all what I would expect...

Peter

 

[Carlos Prio]

Alexa is an extremely competent platform from an old established brand. It will further allow the "old dogs" to make the transition to digital while keeping their self-percieved high-end status. We on the other hand will continue to pioneer & blaze new paths into progressive imagery with new hyper capable tools that still fall into an established commercial workflow (unlike 5d/7d)- Red One & Epic!!!

All that being said... Jim, please make sure Epic f's all others!!!!!

Patiently sitting on the Red Train,
Prio

 

[David Battistella]

Thanks for letting the lid off a bit Jim. LOL

David

 

[Graeme Nattress]

With your recent involvement with DSC, are you guys looking to produce a 5k or 6k chart?

Yes - we're progressing with a high resolution chart at the moment.

Graeme

 

[Mohammed El Sharqawy]

Alexa resolving to only 1.6K from a 3.5K sensor and a 2.88K RAW stream. I am with you there, Peter. Arri has some explaining to do.

this is the question burning inside my mind... HOW?
3.5K sensor -> 2.88K RAW is this line skipping? or Regioning?
2.88K -> 1920 what kind of scaling algorithm that made it only resolve 1.6K??
I think there is something we don't get about this camera..
will scarlet face the same problem with resolution when downscaled to 1080p?? coz of the non uniform scaling?

 

[Jose Lomeña]

Have you done it? On both cameras? It's all hearsay and assumptions based on slippy ground.

Hans

My assumptions are between small and big noise pixels. Small is better for noise reduction, at least this is my experience. If alexa has more or less noise than epic, I don't know that because i didn't make this test.

Saludos,

Jose.

 

[Graeme Nattress]

this is the question burning inside my mind... HOW?
3.5K sensor -> 2.88K RAW is this line skipping? or Regioning?
2.88K -> 1920 what kind of scaling algorithm that made it only resolve 1.6K??
I think there is something we don't get about this camera..
will scarlet face the same problem with resolution when downscaled to 1080p?? coz of the non uniform scaling?

3.5k is for look-around, the 2.8k area is, we assume, demosaiced and downsampled to HD resolution with a 3:2 fixed scaler. This is very similar to how the D21 worked.

Graeme

 

[Kemalettin Sert]

which is more important for Red,DR or Resolution?

 

[Roberto Lequeux]

Personally I'm not too much into the resolution is all thing.

Who is saying it is all about resolution? Not even Jim says that, who owns a company who's future partially depends on 4k being the future. He says 4k taking hold will make RED even more relevant, even more successful, and he is certainly betting on that... but he never said that IQ is all about resolution. You are simplifying.

Just look at what he is doing with his sensor program. What was Mysterium rated at? Everyone gave different numbers, but let's call it 9.5 stops and avoid thread derailment. Then comes in M-X, at 13.5 stops. And they already have a live Beta of Mostro, which is supposed to speed past M-X by a stop, or TWO. Or just look at Mostro closely, instead of packing insane resolutions into a FF35 sensor they are going for larger photosites and "only" 6k. It speaks for itself.

To me RED is clearly aiming at having everything above 4k resolved, then their resolution "hype" becomes just that, misinformation.

RED clearly values DR, like there is no tomorrow! DR and color make your information accurate. You can have a 50k image but if your colors are all wrong, or your DR is minuscule, then you don't have more information than a SD camera with great color and wide DR. They know all of this, all too well. That is why the images from their first generation sensor were so revolutionary, and why the images from their second generation sensor (still in a first generation body!) are having such a profound impact on the industry... just look at the exponentially growing list of blockbusters that are trusting their gargantuan budgets to RED.

There is no "RED fanboy", there is people with eyes. A lot of them like RED IQ better. This whole fanboy thing is getting kinda old... even by my poud-fan standards.

 

[Álex Montoya]

Hans, you should use your real name.

 

[Jay Gannon]

It is his real name?

 

[Timothy Eaton]

Additional note... 5K (5120x2560) is approximately 6.5 time the resolution of 1080P. If you take measured resolution of 4K vs. 1.65K, it is approximately 5.5 times the resolution.

Jim

For reference sake, what is the theoretical resolution ratio (based on target size) between 16mm and 35mm film. Seems like it could be comparable to the difference between Epic and Alexa.

 

[Graeme Nattress]

which is more important for Red,DR or Resolution?

Both are important image parameters.

Graeme

 

[Roberto Lequeux]

Really, the great shocker here is the ~1.6K resolution of Alexa. I would like to see an official chart from ARRI, given the fact that their sensor is actually 3K downscaled to HD. And NO - I am not questioning Jim's test at all. I just want to give ARRI the benefit of doubt and the opportunity to elaborate, if indeed this is the Alexa's resolution...

Peter

Agreed!

this is the question burning inside my mind... HOW?
3.5K sensor -> 2.88K RAW is this line skipping? or Regioning?
2.88K -> 1920 what kind of scaling algorithm that made it only resolve 1.6K??
I think there is something we don't get about this camera..
will scarlet face the same problem with resolution when downscaled to 1080p?? coz of the non uniform scaling?

I don't know enough about the subject to comment when the likes of Graeme step into the room. But I can guess on a different front, a non-technical front.

Perhaps, ARRI might think they need to edge M-X at least on one of the most important fronts, even if by a hair. DR could be argued as the single most important spec at this time. Certainly, in the middle of this race to match film's DR it does get the most word of mouth, or close at least.

And then, perhaps all this downsampling is necessary to get a clean enough image to do so.

Then sticking to a downplaying the resolution leap Jim and Co. have achieved might be the best strategy for ARRI.

Bottom line, if this difference were to be clearly under a 1 stop, and what I get on the trade-off is more than double the FPS, about 1/4 the minimum size and weight, smaller files, and way higher resolution, etc, etc, etc, the list goes on and on... Then you start to think, can I capture it with 13.5? Or do I reeeealy need 13.(8?) stops for the project?

 

[Michael Olsen]

It's interesting to be sure! Thanks for posting the shots, Jim.

As others have noted, resolution is creating somewhat of a secondary effect that can be noticed in the DR tests. Even though the Alexa does show more DR (props to ARRI), the larger pixel pitch seems to come along with slightly chunkier noise, whereas REDs much higher pixel count and smaller pixel pitch leads to a much tighter noise pattern (which seems somewhat more apparent to me on the image). It's almost like 16mm compared to 35mm. Of course, this observation is coming off a wee little JPEG posted on a forum and viewed on an iMac. It could certainly look different at a 200% crop TIFF on a reference monitor.

While an Alexa isn't in my plan simply due to budgetary reasons, I hope to have the opportunity to shoot with one regardless.

EDIT: Also, does anyone remember the ISO settings for each camera used on the test?

 

[David Rasberry]

For 16:9 HD output I think the Alexa is shooting an active pixel region of 2880 x 1620, not the full 3.5k. This is 1.5 x the horizontal and vertical resolution of HD and makes for cleaner scaling algorithms. Why they are not resolving to near the Nyquist limits of 1080p on par with an F35 is a valid question though.

 

[Shane Betts]

After the fuss of today's RC thread, it's good to see Jim and co chiming in with something a little more objective. I wish there was a forum where both companies could post the results of factual, properly carried out tests (as I believe these have been) rather than some of the PR-driven drivel that's flying around.

For me I see nothing in the Epic image to make me consider another camera, given the price and weight considerations. Of course time and extensive shooting of real pictures will tell - and I can't wait for those who will no doubt have theirs way before we get ours to show the images. But let's keep it balanced, please.

 

[Stuart English]

16:9 HD

16:9 HD

For 16:9 HD output I think the Alexa is shooting an active pixel region of 2880 x 1620, not the full 3.5k.

Sounds reasonable. When shooting for 16:9 HD, this 2880 x 1620 RAW pixel count is what Scarlet 2/3" uses... for similar ease of scaling reasons.