Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Testing EPIC-M and ALEXA - suggestions?

Steve Sherrick said:
Although from a purist perspective, I agree with you that it might be a better comparison to see the best of what the two cameras can offer, the reality is that a lot of productions are using the Alexa in Log-C ProRes 444 to the SxS cards. So, if I were doing a test I'd actually want to shoot multiple recording modes. ArriRAW uncompressed, Loc-C ProRes 444 and HQ, etc. And on Epic, I would shoot multiple compression ratios as well for comparison. And make sure to keep really detailed notes along the way. It's time consuming to do a good, thorough test. And the test needs to be for You, as any test will always be questioned if people weren't there first hand to witness it. :-)
Click to expand...

That's a test I've already done with Alexa footage more than once. The results, perhaps surprising to some, are basically that the images in terms of color and dynamic range are essentially identical. You can even apply exactly the same grade on the images and get an identical result. The difference is primarily in sharpness and detail, with the edge going to (no surprise) either a 3K debayer or a 2K frame derived from the 3K debayer. However, it should also be said that the difference is really only evident on a large projection screen. When downconverted to HD, and viewed in that format on a typical monitor, there is very little difference, even with the ProRes compression. LogC does a very good job at retaining just about all of the useful information, just like RedlogFilm does. While it is certainly valuable at times to be able to vary the exposure setting as a post process (an advantage for Red), the fact is that all grading is ultimately performed on an RGB frame, so what is handed over to the grading system is what's important. There are a number of features recording in ArriRaw from the Alexa, and based on my own experience, this is a good thing in terms of resolution and can certainly be seen on a big screen.

Now it should also be said that this is not a commentary on Epic, which by its nature yields a higher resolution image and has a number of other advantages (overcrank speeds, for one, and size and weight, for another). But Steve was commenting on Alexa's multiple recording options and the differences in the results obtained by using them, so I thought I'd offer my perspective based on some direct experience with them.
 
With our internal tests against the two....(As we were thinking of buying a couple of Alexia's at one stage) The Alexia shooting PRORES was a complete non event against a RED's R3D's.
When we down sized to PRORES 2k (REDLOGFILM) to work the same as Alexia's PRO RES, Grading both PRORES files side by side, REDLOGFILM PRORES had alot more range, and we were able to not have a burnt in colour gamma compared to the Alexias PRO RES. So that made it easier to get a good look quick.
RAW side by side, I felt they were quite similar in latitude and range viewing on a HD monitor, but the RED was sharper and cleaner I think due to the resolution difference, and the working in the R3D's was a lot quicker (using our internal workflow).

Now... that was a REDONE MX.. Against the Alexia.
To match an EPIC against the Alexia, I feel is a bit of a unmatched fight. But still very interested to see the outcome.

Super low light tests really really show the imperfections, especially when trying to retrieve information from both. Just make sure you use the same lens (Swapped between cameras) , pointing in the same direction, same Fstop, same camera ISO.

Happy testing :)
 
Mark Toia said:
With our internal tests against the two....(As we were thinking of buying a couple of Alexia's at one stage) The Alexia shooting PRORES was a complete non event against a RED's R3D's.
When we down sized to PRORES 2k (REDLOGFILM) to work the same as Alexia's PRO RES, Grading both PRORES files side by side, REDLOGFILM PRORES had alot more range, and we were able to not have a burnt in colour gamma compared to the Alexias PRO RES. So that made it easier to get a good look quick.
RAW side by side, I felt they were quite similar in latitude and range viewing on a HD monitor, but the RED was sharper and cleaner I think due to the resolution difference, and the working in the R3D's was a lot quicker (using our internal workflow).
Click to expand...

It sounds like you compared Alexa images with the Rec709 LUT baked in to Red images converted through RedlogFilm. I would suggest that this is not a sensible comparison. You really need to use Alexa images that were recorded with the LogC gamma LUT for that kind of comparison.
 
If we're talking personal observations, then what immediately struck me, while working with C-Log Prores files are just how much those images looked like a film scan. I doubt, it is even possible to tell which one is which. On the other hand, Epic has a bit colder, cleaner almost clinical look. Other than that, personally, they are very close. That is, if you don't count superior resolution and dynamic range with HDRx.
 
Mark Toia said:
With our internal tests against the two....(As we were thinking of buying a couple of Alexia's at one stage) The Alexia shooting PRORES was a complete non event against a RED's R3D's.
When we down sized to PRORES 2k (REDLOGFILM) to work the same as Alexia's PRO RES, Grading both PRORES files side by side, REDLOGFILM PRORES had alot more range, and we were able to not have a burnt in colour gamma compared to the Alexias PRO RES. So that made it easier to get a good look quick.
RAW side by side, I felt they were quite similar in latitude and range viewing on a HD monitor, but the RED was sharper and cleaner I think due to the resolution difference, and the working in the R3D's was a lot quicker (using our internal workflow).

Now... that was a REDONE MX.. Against the Alexia.
To match an EPIC against the Alexia, I feel is a bit of a unmatched fight. But still very interested to see the outcome.

Super low light tests really really show the imperfections, especially when trying to retrieve information from both. Just make sure you use the same lens (Swapped between cameras) , pointing in the same direction, same Fstop, same camera ISO.

Happy testing :)
Click to expand...

Mark,

Yeah, I got a similar impression when I had a chance to play with the Alexa. The Alexa is certainly probably the best of the HD crop of cameras now, simple to use, beautiful HD images, sturdy and of course an ARRI. That last one means a lot to many people. But it still ain't an Epic, that's for sure. Not by a long shot.

I find it a bit disingenuous that they were more or less bagging 4K, then behind the scenes developing their own 4K camera technology. As the rumor mill goes. But what do I know! :)

By the way, Mark. I just had a look at your reel linked from your sig and noticed that you had a gig with Petronas Oil. I recently DP-ed the Tokyo portion of their recent F1 promotion. Which crew did you work with from Malaysia? Not the ThinkTank boys perchance?

Cheers

Ivan
 
jake blackstone said:
If we're talking personal observations, then what immediately struck me, while working with C-Log Prores files are just how much those images looked like a film scan. I doubt, it is even possible to tell which one is which. On the other hand, Epic has a bit colder, cleaner almost clinical look. Other than that, personally, they are very close. That is, if you don't count superior resolution and dynamic range with HDRx.
Click to expand...
Great thing about RED in this regard is that it doesn't have to look that way. For those DPs who want to "bake in" a softer, more organic look there are ways to achieve this at the lens. I'm always happy to see DPs make those decisions with the RED. They tend to get some really amazing looks. Certainly there are times when leaving it clean is the more appropriate way to go, but not always.
 
I'd look closely at different compression ratios on the epic. The amount of compression artifacts I've found in the blue channel above 5:1 were quite surprising to me.
Shoot a nicely lit medium close up of a person on each camera besides technical tests.
Shoot a challenging contrast scene on each - wide - medium - close and try to match them in post.
You will find that both are very close in performance one sharper and more resolving and the other softer and more forgiving in extreme highlight/lowlight/contrast conditions. Rejoice that there are creative choices to be made on every job to come....
 
Thanks Jim. Brent clued me into that. My bad (or my dyslexia.)
 
Jannard said:
And if you feel the need to match the highlight softness of an Alexa, just add a .5 low con filter to the EPIC. It seems to be a close match to the OLPF of the Alexa. Just with lots more resolution.

Jim
Click to expand...

This sounds about right. I dunno how much light the .5 lowcon takes, but I might have added an ND6, too if I wanted to shoot at the same stop.

It makes a big diference on the Alexa whether you on the one hand is shooting uncompressed/prores 444 RGB (almost identical) or RAW.

It makes a huge difference whether you're using the in-cam debayering or an external one on the Alexa.
It's a bit like taking off the vaseline going RAW.

The resolution holds up much better, too...

Generally, though I think you'll find the most interesting differences on both cams in the extremes of the exposurerange and if you compare controlled vs uncontrolled light situations (but push them....)
 
Shoot some behind the scenes footage so wee can see what you are doing side by side. Please:-)
 
Thanks for all the input/suggestions everyone. Keep em coming as it looks like we have some time - we are going to postpone a bit as we'd like to put 31974 build on the EPIC.
 
I would love to see Tungsten interior, nicely mixed with some daylight (blue hour) Exposed well and the n 2stop´s under - Candels - And for the pefect senario - around an open fire were you use the light from the fire (and color) as you´r key light..

Use filter´s to achive your image as you want it in the camera - Looking forward for the result..

What lenses are you planning to shoot on ??
 
EPIC vs ALEXA testing suggestion

EPIC vs ALEXA testing suggestion

I would suggest the following test:

  1. FPS: frame-rates of 60fps and 120fps, while panning quickly.
  2. Location: standing inside a dim tunnel and panning from the inside wall of the tunnel to the bright opening at the tunnel's entrance (at noon on a sunny day).
    You should situate a person next to the tunnel's wall, and another person just outside of the tunnel's entrance.
  3. Recording media: SxS PRO cards on the Alexa, 128GB/256GB RED SSD on the EPIC.
  4. ISO/gain: Do the same panning motion on both cameras with ISO set to 100, 1600, 3200, 6400.
  5. Resolution: Since the Alexa can only do 2880 x 1620, which is roughly 3K 16:9, you should set the EPIC to 3K for HD, which is the same resolution.
  6. Recording format: R3D on EPIC (and then convert to ProRes 422 HQ in post); ProRes 422 HQ on ALEXA (see note below). Do not use an external recorder to record ARRI-RAW.
  7. De-bayer the footage from both cameras to 2K 16:9. I'm not sure if you need to de-bayer the ProRes 422 HQ from the Alexa (If not, use a good down-res method to bring it down to 2K 16:9).
  8. Do some shadow/highlight enhancement in post (until you're satisfied from each footage), on a very high-quality and calibrated monitor (at least 27" diagonal, and at least 2560x1440 resolution).
  9. Project at 2K 16:9 on a large screen (300" diagonal should be fine).
  10. Upload to Vimeo at 1920x1080, so we can all evaluate your testing.


NOTE: ARRI recently announced a High-Speed mode for the Alexa of up to 120fps using ProRes 422 HQ:
http://www.arri.de/news.html?article=685&cHash=40b8a55599
 
Last edited:
Anyone else confused by #5 on Eric's list?
 
You can set the EPIC to any of the following resolutions:

5K FF: 5120 x 2700 px (1.896:1) = 13,824,000
5K 2:1: 5120 x 2560 px (2:1) = 13,107,200
5K HD: 4800 x 2700 = 12,960,000
5K WS : 5120 x 2160 px (2.37:1) = 11,059,200

4K 16:9: 4096 x 2304 = 9,437,184
4.5 Wide: 4480 x 1920 = 8,601,600
4K 2:1: 4096 x 2048 = 8,388,608
4K HD: 3840 x 2160 = 8,294,400
4K 2.4:1: 4096 x 1706 = 6,987,776

3K 16:9: 3072 x 1728 = 5,308,416
3K 2:1: 3072 x 1536 = 4,718,592
3K for HD:2880 x 1620 = 4,665,600
3K 2.4:1: 3072 x 1280 = 3,932,160

2K 16:9: 2048 x 1152 = 2,359,296
2K 2:1: 2048 x 1024 = 2,097,152
1080 HD: 1920 x 1080 = 2,073,600
2K 2.4:1: 2048 x 853 = 1,746,944
 
I understand you can set it to any resolution, but isn't the point to see the differrence when they are both operating at optimum quality. Why handicap the test?
 
Clint, the EPIC has on-board RAW recording capability, and a sensor that can go as high as 5K. It wouldn't be fair to compare 5K FF to Alexa's 3K footage.
You need to put them in some sort of equal format, and then compare them.
 
Eric Z said:
Clint, the EPIC has on-board RAW recording capability, and a sensor that can go as high as 5K. It wouldn't be fair to compare 5K FF to Alexa's 3K footage.
You need to put them in some sort of equal format, and then compare them.
Click to expand...

It won't be an equal format, as you'll be blowing up the shot on the Epic, in terms of field of view, and abusing the limited optical resolution of the lens, presuming you're using the same lens on both cameras. Just to match sensor pixel numbers. You'll be comparing apples and oranges, in all regards, except sensor pixel numbers, and be a useless test.
 
You must log in or register to reply here.
Back
Top