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 :)

Epic Compression Test 2

A

Adam Rosenbloom

Active member
Joined
Oct 12, 2011
Messages
29
Reaction score
0
Points
0
Location
Washington, D.C.
I did a new compression test on our Epic today after updating to the latest release firmware (3.0.7) and doing a blackshade calibration. If you're unaware, I did a compression test about a month ago and posted my results on this board. I also sent the results to RED, and they asked me to upgrade (at the time, 3.0.7 was in beta, so I was on 3.0) and recalibrate. I also had a request from members of this board to include a pan to see how the compression held up in motion.

So, I recreated the original test as best I could, picking a day with the same weather/lighting conditions, and using all of the same settings and setup. The only difference between the two tests is the firmware. Additionally, I shot a pan, and those R3Ds are taken from about halfway through the pan.

Here are the results:
http://www.filesanywhere.com/fs/v.aspx?v=8a70658c5f6570aca8

Now, here are the results from my original test (which did not have a pan):
https://www.filesanywhere.com/fs/v.a...8b5d6374b6a5ad

My conclusion last time was that, on difficult scenes, 3:1 was great, 5:1 was good enough, and after that blocky compression started to appear. This second test seems to uphold that conclusion. It just looks about the same to me! I guess that's a good thing; I wouldn't expect or want a wide variance in picture quality from firmware updates. There seems to be a slight variation in the saturation, but if you set the white balance to default that should line it back up.

Here is more info on my setup -

Scene: Isolated shot of trees over a street, including a large foreground tree, where the bark really challenges compression. The weather was overcast, same as last time.
Lens: Same lens, a Zeiss cp.2 35mm lens, with a small aperture of about f/15 to f/16 to get a sharp image that would really test the compression. I did not use an ND filter.
Camera Settings: Same settings. The ISO was 640. The frame size was 4K HD. The framerate was 29.97 and I used a 180 degree shutter angle. All other settings that I can think of were set to their defaults (except white balance).
Notes: This is a rough test to put a camera through, but we may shoot scenes like this in the future.

I would love to hear any of your analysis. Does it look better to you? Worse? Did I include enough information?

Thank you again in advance!

Adam
 
Thanks men!
 
It was my pleasure!

What I find interesting is that 3:1 did perform noticeably better than 5:1, even though 5:1 is supposedly the sweet spot and many films are shot at that compression.
 
The Law of diminishing returns has leaned towards compression of 5:1 as the more favorable compression to choose.

The cost factor in size of drives, length of cards, and over all cost of going to 3:1 does not seem to be worth all the extras for the minimal gains you get in return.
Even though on some complex scenes that can benefit from lower compression, It is more rare and difficult to see the overall value in such a higher expense when it comes to the overall budget.

So dont just base going to 3:1 on this 1 test.
Test for yourself in a real shoot situation and pay attention to how much hard drive space you will be investing in the big picture.

Dont get me wrong I shoot 3:1 and love it
however that was for an IMAX 3D feature.
It has its place and usefulness.


All the best
Luis
 
just a quick FYI, most lenses performance drop off when stopped down past 8.0. Usually top performance is between one and three stops down from open, then sharpness suffers along with other issues.

thanks for the tests, very helpful.
 
It definitely is a cost/benefit analysis with compression. I imagine a cinematographer on a feature shooting 5:1 might go to 3:1 for more complex high contrast scenes.

There was some debate about the effect of the small aperture on the image in my original test, but for consistency, I had to repeat the test in the same setup. The real purpose of this second test was to compare the results after upgrading the firmware. RED didn't really comment on the test, except to ask that I upgrade.
 
After sending my latest results to RED, they finally told me that my results are to be expected, which seems pretty satisfying. I hope this test is a useful reference for others!

Thank you all for your help and analysis,

Adam
 
Hello,

Thanks for your test.

I've allways been shooting 6:1 and 8:1 compression ratio, as I used a lot of shallow depth of field shots and had never had any compression artefact.

Your shot is incredibly usefull to understand how high contrast and variable details from lowlight details (tree trunk) to high lights details (house roof) affects the compression results.

Thanks again.


Patrick
 
Huge thank you ... also for providing all the files. It is real interesting looking at the size of the files, and it's really the DI files that get big, and the total size between the rc3 and rc5 isn't that much compared to DI. I shoot a lot of paintings and vegetables, so the shots that have the tree bark are very applicable to what I'm doing. There is a huge haptic loss of the bark in your tests - between rc3 and rc5 when viewed on a 2.5k monitory. (i.e. in a shot of vegetable, you want the viewer's first thought to be to pick up the vegetable, and not just look at it). I keep on moving my movie spec's RC down to a lower number, I've been able to reconstruct the detail from a rc5, but it takes a lot of "thought" to do that. I think I'm going to bite the bullet and just shooting at RC3 from here on out (raw and compression never has made much sense to my brain anyway).

[I'm not sure, but btw looking forward to the 3D Jerusalem movie in 2013, I am assuming your working with Scott Hoffman who is the operator of the street level RED 3D rig ... [edit - oops ... was wrong ]
 
Last edited:
Did an exterior interview shoot this week with lots of leaves and grass with a hand held Epic . My take away when looking at the footage shot at 8:1 is that RC breaks down big time when presented with complex textures and with everything moving. Next time I'm shooting 3:1 and, if that doesn't work, using a different camera.
 
I guess this explains why shots done at f1.2 on Red one sometimes somehow looked sharper - because there was blur in most of the frame, so the compressor could do a better job of the remaining sharp pixels.

If 3:1 means people won't be able to firehose shots all day without thought, editors everywhere will be very grateful to Epic's compression issues.
 
PatrickFaith said:
[I'm not sure, but btw looking forward to the 3D Jerusalem movie in 2013, I am assuming your working with Scott Hoffman who is the operator of the street level RED 3D rig]
Click to expand...

I think this comment was directed at me, but I'm afraid you've got the wrong man! Sorry! Sounds like an interesting project, but I'm not shooting any 3D in Jerusalem - that I know of, anyhow.
 
Adam Rosenbloom said:
I think this comment was directed at me, but I'm afraid you've got the wrong man! Sorry! Sounds like an interesting project, but I'm not shooting any 3D in Jerusalem - that I know of, anyhow.
Click to expand...
I must have brain farted ... was probably reading something biglu was saying ... then got distracted ... continued reading ... Can I blame it on a compression error in my brain ?

Max for the scarlet at 4k is 6:1, so I guess my decision is made for me. In some ways the epic has so many options, it boggles my mind, is kind of nice for people doing their first feature to have a reduced set of options, forces a bit of focus on "normal" configurations.
 
Last edited:
PatrickFaith said:
I must have brain farted ... was probably reading something biglu was saying ... then got distracted ... continued reading ... Can I blame it on a compression error in my brain ?

Max for the scarlet at 4k is 6:1, so I guess my decision is made for me. In some ways the epic has so many options, it boggles my mind, is kind of nice for people doing their first feature to have a reduced set of options, forces a bit of focus on "normal" configurations.
Click to expand...

No worries, brain farts are common where I'm from!

More options can be good and bad. If you have the time, more is better! Half the time I'm an editor here, though, so it can be a challenge to get all of the tests done.
 
Thanks for doing this!
Oddly though, I shot something yesterday at 120fps and 12-1 compression and 5k (I know you're shooting 4k HD), with a similar high contrast forest scene, and it looked like the results from your 5-1... MUCH better than what your 10-1 looked like. Not sure why this is. Maybe the 5k?

EDIT: Taking another look at my footage, there was also a bit less contrast, but not a whole lot less than yours.
 
I think for complex scenes like these, lower rezz and no HDRx may be of advantage, in addition to lower compression...

Just a hunch...
 
Thanks for posting this test. I came across several challenging complex nature scenes that are unusable at 8:1 compression on Epic. Strangely I don't remember such scenes to be challenge for RED One (mysterium sensor).
 
You must log in or register to reply here.
Back
Top