We've relaunched as a full community! Get the scoop:

March 6, 2013

Blackmagic Cinema Camera, RED EPIC, and Arri Alexa RAW Camera Test Part 3: Overexposure

This is a guest post by Cinematographer Ryan E. Walters.

Welcome to Part 03 of Some Like It RAW, where I am comparing the Arri AlexaBlackmagic Cinema Camera, and the RED EPIC. My goal for these tests is to explore how each of these cameras handle real world shooting environments. Part 01 explored how these cameras handle IR pollution. In Part 02, I tested underexposure. And here in Part 03, I'm exploring the world of overexposure and diffusion filtration. Continue on to watch the 10 minute video, read my summary, and get the downloadable RAW frames from each camera.

I'd again like to thank the following people who helped to make this test possible:

Picture This Productions
(Provided the Blackmagic Cinema Camera)
Shawn Nelson
(Provided his EPIC)
Patrick Eggert
(Provided his Alexa)
Isaac Marchionna
(Assisted for the shoot)
Laurie Slater
(Was our Talent for the day)
180 Films
(Provided the Cooke 20-100mm Zoom)
Letus
(Provided the cage for the Blackmagic Cinema Camera)
Schneider Optics
(Provided the IR Tuner Kit)

As I have stated in the previous two tests, my goal in this series is to stress and break these cameras to see how they perform in their extremes. It is not to prove that they can create beautiful images -- that has been done. I am also not interested in furthering the religion of any one of these cameras. So if you are looking for material to support your beliefs, or camera purchase, then this test will disappoint you. But if you are interested in seeing how these cameras perform so that you can learn where and how they break in order to push your own creativity, then continue on.

Here is the overhead diagram of how I lit this scene. For the close up I added a frame of 4x4 full grid and left the light in the same position.

Below the 10 minute video you will find details of how I determined the overexposure values, what I think the dynamic range of each cameras is, exposure recommendations, a summary of my observations, and a link to the downloadable RAW frames from each camera.

How I Determined Overexposure Values:

I ran two tests to determine the overexposure values for these cameras. For the first test, I took the Sekonic Profile Target II exposures and loaded them into the Lite version of Resolve. Once there, I pulled up the waveform and looked for the first chip that had reached the clip point. (The point at which the line completely flattens out). This value gave me the maximum overexposure value, or the clip point. Then I counted down a couple of chips to a position where I felt the waveform had enough detail in it that I could record some discernible highlight detail. This is the value that I used for my safe, or conservative readings.

Waveform of Sekonic Chip Chart

Close Up of Waveform of Sekonic Chip Chart

For the second test, I loaded the files into Photoshop, and I selected each patch individually to acquire its Median value. As soon as this number started to decrease in value, I knew I had found the point at which I had found the clip point. These values lined up with the waveform results -- which isn't surprising to me, as it is much easier to determine an absolute clip point in digital than it is to determine an underexposure value.

Selected chip of a +8 EV exposed Sekonic chart in Photoshop

Results of Selection

Why didn't you use a DSC Xyla Chart?

Simple -- I don't own one, and it is outside the budget I am going to spend on a testing chart. I recognize that the Xyla is THE standard for testing dynamic range -- no arguments here. But for practical purposes, I have found the $130 Sekonic Profile Target II chart to suit my needs and applications when it comes to creating camera profiles for my meter. If you are feeling generous and want to donate the Xyla-21 to me, I will gladly use it in future testing :)

Why do you have a safe or conservative value for overexposure?

Although the clip point is an absolute value that it easy to measure, I am not a fan of clipping at the camera level in digital cinema. By being more conservative in my overexposure values, I am allowing for a touch more headroom for highlight roll off. When it comes time to grade the footage, I have found that this small amount of headroom allows for a smoother transition when it is pushed in the grade. If it is all about the numbers for you, then you can ignore my conservative numbers, as they are just my personal preference :)

Why didn't you use HDR-X on the EPIC?

HDR-X is a great way to extend the dynamic range of the camera. And although it will allow me to record more information in the highlight, I don't like the way it renders motion. It is all about personal preference. If you like HDR-X and use it in your shooting, then you can extend the overexposure values I have listed by your preferred HDR-X settings.

What about the RED Dragon Sensor and the Alexa XT?

I'm interested in testing cameras I can actually use and shoot with today, not years from now. When RED finally releases their new sensor and it is available to shoot on by us mere mortals, I'm sure it will be great. (I am looking forward to its extended dynamic range). But in the meantime, I need to explore and use tools that are available to me now. The same goes for the Alexa XT. Here in my market, the Alexa is in high demand and is very rarely seen sitting around the rental shop. That makes me wonder how quickly they will be upgraded.

Blackmagic continues to be slow in shipping their cameras and while they are still not widely available to the public (as of this writing), I can go to my local rental shop and use it today. (Which is what I will have to do until my MFT version finally arrives...).

Observations & Recommendations:

  • Smaller screens make it harder to see detail in the highlights, while bigger screens allow for the opportunity to see detail. Nuances in-between the cameras get lost the smaller the viewing screen. I had a harder time telling the cameras apart on my iPhone than I did on my 50" Plasma.
  • Combining overexposure and diffusion filtration is a great way to create a glow, or dreamy look in camera. (If you want to see more filter tests, check out this Filter Test I did).
  • Diffusion can help to smooth the transition to overexposure. It does not extend the dynamic range of the camera, but the blooming of the highlights does make the roll off more graceful.
  • All of these cameras are more than capable of creating beautiful imagery. If you are consistently getting bad results from any of them, it is not the camera that is causing the problem...

Arri Alexa:

  • Total overexposure = 7 2/3
  • Conservative overexposure = 7 1/3
  • The extended amount of range in overexposure allows the Alexa to have the most graceful roll-off of the three cameras. This is why it has been one of my preferred cameras to shoot with to date.
  • Older lenses look great on it, as the imperfections in the lens add a more organic feel to the imagery without degrading it too much.
  • This camera doesn't need diffusion filtration, as the images do not feel overly sharp. The choice to use diffusion will be more about making the talent look good than it will be about removing an artifact from the camera.

Dynamic Range (DR) at EI 800 As Profiled In Log Using the Sekonic Profile Target II [8-Bit Photoshop values]

  • Overexposure (Total) = 7 2/3 [243]
  • Overexposure (Safe) = 7 1/3 [235]
  • At Exposure = 0 [106]
  • Underexposure (Safe) = - 5 5/6 [31]
  • Underexposure (Total) = - 8 5/6 [22]
  • Total Measurable DR = 16 1/3
  • Usable DR = 13 1/6

Exposure Recommendations Based on DR: (YMMV)

  • I have had the best results when exposing the camera at EI 800 by keeping all important highlight information within 7 stops, and all important underexposure values within 6 stops. Any detail that is recorded beyond these levels is a bonus. As the EI decreases in this camera, the image gets cleaner, and there is less headroom in the highlights. As the EI increases in this camera, the image gets noisier and there is more headroom in the highlights.

Blackmagic Cinema Camera:

  • Total overexposure = 5 1/6
  • Conservative overexposure = 4 5/6
  • It transitions quickly into overexposure, and the overexposure tends toward yellow. So carefully choose what you allow to clip.
  • Diffusion is this camera's friend. The lack of an OLPF in this camera causes it to have overly sharp images which is not flattering -- especially on a close up. By using diffusion on the camera you can make the image more forgiving on the talent.
  • Stay away from Rec709 for best results, and make sure that the images are handled correctly in post. I shot this test in Film Mode but when the files come into Resolve they show up as Rec709. The Rec709 images looksWAY overexposed. (And made my heart skip a beat, before I realized what was going on). If the files are processed without being changed to BMD Film first, you will NOT be able to get the best imagery out of the camera.

Dynamic Range (DR) at EI 800 As Profiled In Log Using the Sekonic Profile Target II [8-Bit Photoshop values]

  • Overexposure (Total) = 5 1/6 [251]
  • Overexposure (Safe) = 4 5/6 [245]
  • At Exposure = 0 [96]
  • Underexposure (Safe) = - 5 [15]
  • Underexposure (Total) = - 7 1/6 [9]
  • Total Measurable DR = 12 1/3
  • Usable DR = 9 5/6

Exposure Recommendations Based on DR: (YMMV)

  • I have not worked with this camera long enough to have any definitive recommendations. Based off of my test results and what I have experienced so far, I recommend that all important highlight information be kept within 5 stops, and all important shadow information be kept within 5 stops. Any detail that is recorded beyond these levels is a bonus. (I have a hunch that there will be some added information in the shadows that will add subtly to the picture). Practically speaking, this means that you will have to expose for your highlights, and then add lighting to compensate for the underexposure areas of the image if needed.

RED EPIC (with MX Sensor without using HDR-X):

  • Total overexposure = 5 1/3
  • Conservative overexposure = 4 2/3
  • It transitions quickly into overexposure, and the overexposure tends toward magenta. So carefully choose what you allow to clip.
  • Older lenses look great on it, as the imperfections in the lens add a more organic feel to the imagery without degrading it too much.
  • This camera does not require diffusion, but it is helpful to use due to the high-resolution that this camera shoots at. If you want to be more forgiving to your talent, then I'd add a touch of diffusion. :)

Dynamic Range (DR) at EI 800 As Profiled In Log Using the Sekonic Profile Target II [8-Bit Photoshop values]

  • Overexposure (Total) = 5 1/3 [237]
  • Overexposure (Safe) = 4 2/3 [228]
  • At Exposure = 0 [100]
  • Underexposure (Safe) = - 5 1/6 [16]
  • Underexposure (Total) = - 8 1/3 [11]
  • Total Measurable DR = 13 2/3
  • Usable DR = 9 5/6

Exposure Recommendations Based on DR: (YMMV)

  • I have had the best results when exposing the camera at EI 800 by keeping all important highlight information within 5 stops, and all important underexposure values within 5 stops. Any detail that is recorded beyond these levels is a bonus. I have often found that there is extra detail beyond the 5 stops of underexposure, which can add additional subtlety and texture to the image. But if I am counting on this added range, or if I need to pull out information in the shadows I find that I am disappointed with the results. (It gets too noisy for my tastes). I have also found that if I protect my highlights with the Epic, I can get a nice roll off into overexposure. This is why I am more conservative with the overexposure values of this camera. As the EI decreases in this camera, the image gets cleaner, and there is less headroom in the highlights. As the EI increases in this camera, the image gets noisier and there is more headroom in the highlights.

Downloadable RAW Frames (87MB): SLIR-Overexposure.zip

What do you think? Do these results surprise you? Will you be using diffusion filtration with the Blackmagic? If so, what will you use? Do you have any exposure recommendations that you have found helpful when shooting with any of these cameras?

This post originally appeared on Ryan’s Blog.


Ryan E. Walters is an award-winning Oregon-based cinematographer. His work has allowed him the opportunity to travel worldwide in the pursuit of telling stories that are visually compelling. His experience includes feature films, documentaries, commercials, and shooting for Comcast, TLC, Oxygen, and the Discovery Channel.

Your Comment

37 Comments

These videos are extremely valuable but, very hard on my attention span. Loooong titles.

March 6, 2013

-1
Reply

Thanks for the critique- I have noted that, and going forward I'll tighten it up more. :) I appreciate the input. :)

March 7, 2013

1
Reply

Extremely useful test. I was very disappointed with BMCC and EPIC...I wasn´t expecting such a huge difference to Alexa...

Thank you

March 7, 2013

0
Reply
Augusto Alves d...

You weren't expecting a big difference between a 100k camera and a 3k camera?

It is astounding that the BMC is anywhere in the same hemisphere.

March 8, 2013

0
Reply
Shanon Fernald

Yet, it is AMAZING that the BMCC even comes close. :)

March 9, 2013

1
Reply

Well done, Ryan. Will you go to NAB next month?

March 6, 2013

0
Reply
Jason

Yep- I'll be at NAB for the whole week. I'll be working at the Letus Booth. :)

March 7, 2013

1
Reply

Tell me if I'm getting this right. I rounded to the nearest .5 just because I'm lazy, but if I did the math correctly then that means:

-The Alexa gets up to 16 stops of dynamic range, but only 13 (roughly) stops that you would use safely.
-The BMC gets up to 12.5 stops of dynamic range, but only 10 (roughly) stops that you would use safely.
-The RED gets up to 13.5 stops of dynamic range, but only 10 (roughly) stops that you would use safely.

This was done using the underexposure limits and usable numbers and combining them with the overexposure total and safe numbers.

If it's true (based on your tests anyway) then the BMC is a hell of a deal based on dynamic range alone!

March 6, 2013

0
Reply
David

Based off of my tests, and my preferences, yes, correct. If you are more comfortable digging information out of the shadows, then you can get a bit more out of each camera. For the money, the Blackmagic is VERY impressive! (As long as you only need the feature set it offers, you'll be fine.)

March 7, 2013

0
Reply

Just wondering: how expensive are those DSC Xyla charts?
I guess it's one of those cases where you sort of expect it to be massively expensive when the company doesn't openly tell you how much it costs...

March 6, 2013

0
Reply
Blah

Last time I checked (Over a year ago) they were around $4,000 ... pretty costly for one chart if you ask me- but they are the standard. :)

March 7, 2013

0
Reply

Ah, okay -- thanks!
That's definitely up there in price, especially when compared with options from Sekonic and Stouffer, etc. Still it's nice to know.

March 8, 2013

0
Reply
Blah

Excellent Ryan, thanks for the test.
Alexa to be a nice camera;)
I try to understand, there are many forums and a lot of testing around the internet, why have 16 Alexa Stop? I have never seen anything of this.
For example, here Alexa and Epic are very close http://provideocoalition.com/freshdv/story/epic_vs_alexa_dynamic_range, why in your test there are 3 stop difference? 13 for Alexa and 10 for Epic. Arri on its website saying 14 stop,RED 13.5 stop, who is wrong?

thanks so much

March 7, 2013

0
Reply
ALEXXXISS

RED has always made their cameras out to be way better than they are. They say the dragon sensor will have 16 stops of DR but I doubt it's gonna have more than the current model Alexa.

March 8, 2013

0
Reply
James

Hey, here's a "fanboy" the video was talking about.

March 8, 2013

0
Reply
James

It has been my experience that Red tends to over state things, and Arri tends to understate things. Additionally, when they both test their cameras, they are using the high end chart to test them, which are more accurate then what I am using. However, while their charts are more technically accurate as they are back lit, I have found that the results I get from the Sekonic chart are adequate for my uses. Transmissive charts (Like the XYLA) are better for technical measurements, especially at high dynamic ranges. Reflective charts are (Like the Sekonic) more limiting / prone to not being as technically accurate. But when it comes time to actually shoot a scene, the camera is recording reflective values as it is the light bouncing off of objects and bouncing into the camera, not transmissive values (light passing through objects.).

So all that to say, in the end both are "right". The technical chart works best for technical values, while I have found that the reflective chart works adequately for real world use and camera profiling. I hope that helps. :)

March 9, 2013

0
Reply

I and a number of ASC cinematographers have done this test.

We all came to the exact same conclusion - ARRI has another 1-1.5 stops of latitude - not 3 stops.

Apply noise reduction, and Epic gets another stop in the darks which Arri cannot (because you can't noise reduce lower res footage as much.)

I think I'll trust my own tests, and those done by lauded ASC members over these.

March 7, 2013

0
Reply
Robert Ruffo

Just here to be negative or do you care to share?

March 7, 2013

0
Reply

Fare enough. :) That is the point of doing testing and figuring things out. :)

March 9, 2013

0
Reply

"not years from now"

Oh snap, boy...

March 7, 2013

0
Reply
Natt

You'll definitely want to stay away from Rec709 when shooting raw on Blackmagic. But if you're shooting Prores or DNxHD for a quick turn around without a grade, Rec709 looks great.

March 7, 2013

0
Reply
LeighO

What Ryan probably doesn't know, is that, the reason why Alexa handles the highlights better is because, the sensor it has a special technology, which is different from the other two cameras, in other words, Alexa has auto HDR on the sensor ACCORDING TO THIS PAGE:
http://www.arri.com/camera/digital_cameras/technology/arri_imaging_techn....

I am sure that The Epic camera in HDRx mode performs the same, if not better, in the same lighting conditions..

March 7, 2013

-1
Reply

Nope, I am aware of that. But Arri's process is different then Red's, and I personally do not like the aesthetics that Red's process produces. If you like it, that is great, use it. :) Technically speaking you are correct that the HDRx does perform better as you can protect more. But do you like the look / motion rendition - that is a personal taste issue ...

March 9, 2013

0
Reply

Sorry but where is the Sony F55 in this?? It has better DR than red, BM easily.

March 7, 2013

0
Reply
Brandon

It was with you - why didn't you show up? ;) It isn't in here as I did not have access to the camera. If you, or anyone else wants to give it to me, or give me the cash to rent it, I'll test it out. Until then, I'll have to wait until I have a project that needs it before I get to test it out. When you do not have unlimited funds to work with, you have to use the tools you have access to. :)

March 9, 2013

0
Reply

" Practically speaking, this means that you will have to expose for your highlights, and then add lighting to compensate for the underexposure areas of the image if needed." - If I understood this correctly..this is exactly opposite what John Brawley and the rest of the BMCC users in the forum suggest to get rhe most of the BMCC's dynamic range. You should exposed for the shadows, let the highlights blow. In post you can get them all back. Expose to the right as they say.

March 8, 2013

0
Reply
Quobetah

It depends on what you are looking at on the monitor. If you are looking at Rec709 on the monitor, then yes, expose to the right, as the camera is holding a lot more detail in the highlights in the RAW file then what you are seeing. If you are looking at RAW, then I would say exposure for your highlights, and adjust from there.

John doesn't quite say ETTR over ETTL. He is saying that you should ETTR until you clip only the non-important highlight information, so that it is getting as much light as possible. At least that is how I understand what he is saying.

Even if you are working in RAW, if an important highlight detail is clipped, you cannot get it back ...

March 9, 2013

0
Reply

It doesn't matter what you're looking at if you're shooting RAW, the Zebras at 100 percent tell you what you're clipping on the sensor. The "Video" setting is just a quick lut to judge your ratios by, the zebras will always tell you what's really going on as far as clipping goes.

March 9, 2013

0
Reply
Kholi

Thanks for another great test Ryan!
After seeing the effect of applying REC709 LUT files to BMCC RAW files in Resolve, in this video: http://www.youtube.com/watch?v=nrHJTXoSseY I made a note to myself to always start here as a base and then adjust the highlights, shadows and color as needed from there . To my eyes the REC709 corrected the color and contrast of the flat BMCC LOG files pretty nicely for a quick global look. The REC 709 LUT did not look "way overexposed" too me as you have stated here. Perhaps this "way overexposed" look with REC709 only happens when the camera is way overexposed?

March 8, 2013

0
Reply

After looking at the clips again, I do see that the highlights are clipped and blacks are being crushed. But I still think this can be a useful starting point. Is there any reason why begining with a REC709 LUT to get "in the ballpark" is a bad move? Some of the lower contrast scenes seem to hold up fine where others with bright sky did not.

March 9, 2013

0
Reply

If it works, great - there is nothing "wrong" with that approach. It is all about end results, and communication. :)

March 9, 2013

0
Reply

That is possible. But take a look at the RAW files I have available for download and try them for yourself. When they are loaded in Resolve they come up in Rec709, and look beyond salvageable to me. But when I go into the RAW settings (Not the LUT settings) and change it back to BMD Film, I see that I have a lot of information to work with, and the image is not lost like I thought it was.

Whether you use Rec709, or not, is not THE important part / issue. THE important part or issue, is making sure that the end results are arriving at a place where you, and the rest of the team are happy. And when working with RAW, there is a BIG potential for things to go wrong or for miscommunication to happen. If I had handed off this files, and expected the metadata to communicate my intentions, the files would have communicated the WRONG intentions as they came up as Rec709. Which is NOT what I was shooting for on set. It seems to me that the Metadata needs some work with Resolve, as what I was shooting on set, should come up as the default in Resolve to help communicate intentions, but that isn't the case ...

March 9, 2013

0
Reply

"When they are loaded in Resolve they come up in Rec709, and look beyond salvageable to me." I thought that once your in RAW, as long as the hilights aren't clipped in the capture, (which I don't think you did) they can be brought back to a desired level no matter what type of lut might be applied? That an LUT or any other CC applied only gets "baked in" when a final media export is made?

March 9, 2013

0
Reply

HI Ryan,
today I downloaded your files test.
I can not understand certain things.
1. I would have liked to see how clipping the ALEXA, 'cause you did not?
2. I wondered if you are left with the gray 18% the same for all three cameras? or you set the aperture and you have only changed optical camera to camera?
3. I took a look also to the RED files, how come you are below about 3 stop threshold of medium gray? be desired?
I have an Alexa (first model) and Epic, unfortunately I have not had access to a BMC, I tried more or less to redo a similar test but properly exposing the two cameras I can not get to your result is better alexa but only of 1-stop about, after doing the post in Resolve.
However as you say the camera to be just a tool;)

March 9, 2013

0
Reply
ALEXISS

No need to test Alexa XT separately as there are no changes to the sensor or main electronics. Its just an upgrade to the Alexa 4:3 camera line. Also the 16:9 image quality is exactly the same as that of the 4:3 sensors.

March 10, 2013

0
Reply
Carl

I really love learning from all what you are doing. I appreciate all your time and effort too.

March 11, 2013

0
Reply
Steve V

If you prefer the "softer" look of Epic over the Blackmagic, then what is the point of having 5K resolution?

April 10, 2013

0
Reply
ernstdante