New Epic-W vs. Alexa Mini test

[jussi rovanpera]

I assume the Arri dual gain architecture is patented by Arri, and that's the reason no one else is using it?

 

[Rakesh Malik]

AFAIK Black Magic also uses a dual gain architecture in its cinema cameras.

 

[paulcurtis]

Funnily enough coming from the Sony side of the fence one of the biggest draws of the Red is the simplicity of shooting. Yet we're making it look to be more complicated than it is really. I think one of the pros and cons of red is that level transparency - change the OLPF for that extra few % of difference (is not an option on any other camera), calibrate at different working temperatures (all hidden in other cameras), turn on HDRx (perhaps done internally in others or not as comprehensive). And then lots of people still struggle with RAW workflow. I think it *seems* more complicated than it really is.

It's education and taking the time to understand your tools.

Still people in the comments on that thread aren't understanding exposure properly.

This whole roll off thing is interesting for those that want that level of geekiness. One thing i noticed from Sony RAW (The DNG side that allows you to actually get at pre-debayered images) is that the different channels are very different in sensitivity meaning that the ability to reconstruct highlights based on non clipping channels makes a difference, a stops worth of highlights that can then be rolled off very easily. This is something that you can only really do pre white balancing. Perhaps the Alexa is doing all of this internally, before spitting out ArriRAW.

And then Aaron's side by side of the womans face. It's chalk and cheese. So why isn't that image the top?

How many people reading that shootout are looking at full res files, and not just the web delivered imagery?

cheers
Paul

 

[Ben Scott]

Absolutely, having been in a Sony wilderness for a couple of years getting good and consistent results in the Red ecosystem couldn't be simpler as long as you actually know the methodology in the first place.

EDIT: But people need to either chill out about this test or post their own ones. The moment the reduser army wades in en masse on a subject it does far more harm than good for the perception of the camera, the workflow and the company.

 

[Sabovic Adis]

Such mapping happens post A to D, and does not effect raw data. It's part of the raw development process. One of the great things with modern digital cinema cameras (rather than video cameras of old) is that you can always get linear light data out.

Graeme

If not asked too much, does the signal, coming from the sensor, get gamma encoded prior quantisation? I mean, electronically, in analogue domain?
Somehow I feel that 16 bits is not enough for 16+ stops of light, especially not for lows, where "the meat" is. Think Weber's fraction.

 

[Hrvoje Simic]

If you're into resolution, it makes no doubt it's chalk and cheese. Concerning colors, color separation, etc., it's also chalk and cheese, in favor of the Alexa. On that full res image, I see the same differences I noticed when I tested my 5D vs my Mamiya 7 film camera. One image is full of hues and subtle tonalities, the other one is more "monochromatic". Red is fighting on the resolution field, whereas Arri favors the colors. Just pick your tool flavor.

Whenever you are judging color performance of a raw recording camera, by watching a final image closer to real life you are judging:

a) Camera raw response
+
b) Mapped in base gamma and gamut (base color science)
+
c) Transformation from base color science into standard gamma and color space.

These are not video cameras and the same approach in interpreting their performance as inherited from past era cannot be applied.

This is not the final camera performance you see on these tests, this is just one of the outcomes from those particular image processing routes.

 

[Elsie N]

Funnily enough coming from the Sony side of the fence one of the biggest draws of the Red is the simplicity of shooting. Yet we're making it look to be more complicated than it is really. I think one of the pros and cons of red is that level transparency - change the OLPF for that extra few % of difference (is not an option on any other camera), calibrate at different working temperatures (all hidden in other cameras), turn on HDRx (perhaps done internally in others or not as comprehensive). And then lots of people still struggle with RAW workflow. I think it *seems* more complicated than it really is.

It's education and taking the time to understand your tools.

Still people in the comments on that thread aren't understanding exposure properly.

This whole roll off thing is interesting for those that want that level of geekiness. One thing i noticed from Sony RAW (The DNG side that allows you to actually get at pre-debayered images) is that the different channels are very different in sensitivity meaning that the ability to reconstruct highlights based on non clipping channels makes a difference, a stops worth of highlights that can then be rolled off very easily. This is something that you can only really do pre white balancing. Perhaps the Alexa is doing all of this internally, before spitting out ArriRAW.

And then Aaron's side by side of the womans face. It's chalk and cheese. So why isn't that image the top?

How many people reading that shootout are looking at full res files, and not just the web delivered imagery?

cheers
Paul

I agree with the above. A simple ETTR setup gives you what's available from the light and putting a LUT of choice in the chain gets you the results you choose. Thousands of professionals are now, or soon will be, shooting like this and generationally, they like choice.

That is, they have the choice to shoot as described above or dive deeper and setup their camera to output an almost ready-to-exhibit 4k or 8k finish.

In the past, professionals were born out of long hours of work with film cameras... apprenticeships if you will. Nowadays, professionals are being separated from the wannabees by owning their own cameras and training on the job at whatever level that may be initially.

IMO, a test that is both true and at the same time flawed like the one in this thread will not matter to those who are further along in the learning curve.

 

[Hrvoje Simic]

I agree with the above. A simple ETTR setup gives you what's available from the light and putting a LUT of choice in the chain gets you the results you choose. Thousands of professionals are now, or soon will be, shooting like this and generationally, they like choice.

A simple ETTR, oversimplified and generalized will lead to underexposed imagery by trying to preserve data in the lightbulb with zero fill light and at the cost of everything else.

And in many times it has lead to those types of outcomes and many blaming their cameras with 2-4 stops underexposed then overly pushed imagery.

 

[Kemalettin Sert]

AFAIK Black Magic also uses a dual gain architecture in its cinema cameras.

 

[Anton Shavlik]

I have no idea what that dude did. Here's 100% scale at UHD, meaning Helium at a 4:1 downscale, Alexa upscaled from 3.4k to UHD.

(to view full scale: right click, open in new tab)

What I did was match exposure, color temp and lifted shadows. The higher contrast of red's default rec709 of course makes it look sharper but you could easily add contrast to arri and get the same.

Otherwise I did full debayer scaled to 4K in resolve. All my settings are in my posts in this thread.

 

[Hrvoje Simic]

 

[paulcurtis]

Hrvoje,

Nice work.

Assume Helium is the lower one? Which i actually prefer, it has more colour in the highlights but subtle about it.

There's not much in it though - lower has greater DOF hence thinking it's Helium!

(am i right? if not then i can edit this post to swap them around and appear clever... )

cheers
Paul

 

[Hrvoje Simic]

Thanks Paul. Hopefully this keeps some wrists from being slit.

 

[Elsie N]

A simple ETTR, oversimplified and generalized will lead to underexposed imagery by trying to preserve data in the lightbulb with zero fill light and at the cost of everything else.

And in many times it has lead to those types of outcomes and many blaming their cameras with 2-4 stops underexposed then overly pushed imagery.

Just how often does a shot use all available dynamic range?

Anyway, if anyone is anal about shooting ETTR, then dive deeper and impress someone...

 

[Hrvoje Simic]

Just how often does a shot use all available dynamic range?

Anyway, if anyone is anal about shooting ETTR, then dive deeper and impress someone...

Just sayin'...

No one in the audience cares about the light bulb or sun reflections but skin tones pushed 3 stops don't look very good on any camera. Expression tells the story, not the frikking light bulb. Tests like these are not meant for contests but to understand how to light and expose properly with each tool. With today's cameras all/most range of most scenes fits in available DR if there is any understanding and usage of light and proper exposure so general dramas with one stop in highlights are blown out of proportions more than the light source out of DR in those scenes.

Instead of worrying about what one doesn't have there is much more potential in understanding and harvesting what one does have.

 

[rand thompson]

Here is Red file only with iso changed from 800 to 1000.

redwidegamutrgb/log3g10 plus ipp2 - low contrast rolloff 4.

[/url]rwgrgb log3g10 ipp2 low roll 4 by rand thompson, on Flickr[/IMG]


redwidegamutrgb/log3g10 plus ipp2 - low contrast rolloff 4 plus grade.

[/url]rwgrgb log3g10 ipp2 low roll 4 plus grade 2 by rand thompson, on Flickr[/IMG]

 

[Homer Dulu]

A simple ETTR, oversimplified and generalized will lead to underexposed imagery by trying to preserve data in the lightbulb with zero fill light and at the cost of everything else.

And in many times it has lead to those types of outcomes and many blaming their cameras with 2-4 stops underexposed then overly pushed imagery.

Just how often does a shot use all available dynamic range?

Anyway, if anyone is anal about shooting ETTR, then dive deeper and impress someone...

Please excuse my ignorance, but until our cameras can capture a crazy high amount of dynamic range (way way more than 16 stops), wouldn't you still need to light properly, regardless if you use ETTR or not? In this example, if you want to retain all the highlight information in the lightbulb but not underexpose the rest of the scene, you add more ambient light (or dim the lightbulb) so you can reduce the contrast/DR? Isn't that also something that you'll see in the goalposts as well and that the DP should be very well aware of?

I don't think there's anything inherently wrong with ETTR itself, but I don't think it should be an excuse to not ensure that the scene is properly lit?

Edit: rephrased last question

 

[Elsie N]

Just sayin'...

No one in the audience cares about the light bulb or sun reflections but skin tones pushed 3 stops don't look very good on any camera. Expression tells the story, not the frikking light bulb. Tests like these are not meant for contests but to understand how to light and expose properly with each tool. With today's cameras all/most range of most scenes fits in available DR if there is any understanding and usage of light and proper exposure so general dramas with one stop in highlights are blown out of proportions more than the light source out of DR in those scenes.

Instead of worrying about what one doesn't have there is much more potential in understanding and harvesting what one does have.

Turns out we have common ground... by getting all of what one DOES have with ETTR, one can then grade to one's preferred look. '-)

 

[Elsie N]

Please excuse my ignorance, but until our cameras can capture a crazy high amount of dynamic range (way way more than 16 stops), wouldn't you still need to light properly, regardless if you use ETTR or not? In this example, if you want to retain all the highlight information in the lightbulb but not underexpose the rest of the scene, you add more ambient light (or dim the lightbulb) so you can reduce the contrast/DR? Isn't that also something that you'll see in the goalposts as well and that the DP should be very well aware of?

I don't think there's anything inherently wrong with ETTR itself, but I don't think it should be an excuse to not light the scene at all?

Not sure where exposing to the right became synonymous with shooting only available light.

Of course you light the scene if and where needed. Then you expose the scene to capture the full range of available dynamic range... that is, you get as much information recorded as possible.

An exception to exposing-TTR is in a dark scene. Then you use the LLO OLPF and expose to get as much info as possible.

 

[Homer Dulu]

Not sure where exposing to the right became synonymous with shooting only available light.

Of course you light the scene if and where needed. Then you expose the scene to capture the full range of available dynamic range... that is, you get as much information recorded as possible.

Sorry Elsie, I rephrased my question while you were writing your reply. But yes, I agree with you there. Not going to get any argument from me.

And Hrvoje, the thought of pushing skin 3 stops gives me the heebie jeebies. I never did that in photography and I sure ain't about to start with video!