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 Results

I really really appreciate Haakon doing this test. This is what makes Reduser worth visiting.

I know this isn't a comprehensive test to address all situations, but just so I understand, it seems like a large portion of the frame (dark area) has very little detail, at least compared to a field of grass or crashing beach waves. Was that done to specifically test noise in the shadows?
 
Hi Julio,

A lot of people asked me for a very low light situation and I tried to create a scene with that in mind. I was hoping to fill the darker areas with more pronounced compression-taxing rain, but didn't have the firepower with me to capture it the way I would have liked. Just one test; we can do more. :) As it stands I have extremely limited options for shooting since the battery modules are not yet available. Once those are released, I will be able to shoot in much more varied conditions. I am, however, pretty impressed with how the footage holds up throughout the compression range; even the blowing hair doesn't completely fall apart at the extreme end of the spectrum.
 
How about some trimmed R3D's??
 
I got it to work after 30 mins leaving the page up thanks. All the shots look great.. I would not go passed 9:1 based on this test.
 
SeanBrown said:
I got it to work after 30 mins leaving the page up thanks. All the shots look great.. I would not go passed 9:1 based on this test.
Click to expand...
I came away with the same conclusion, though I set my limit at 10:1 (which you can't see in this clip since I only included the odd steps).

I'll try to get some R3Ds posted later that you can push and pull to your heart's content. :)
 
Sam Taylor said:
Neither for me unfortunately. Thanks for uploading though! Would value someone giving a synopsis of the shot results, for us guys with viewing issues. Thanks!
Click to expand...

Just click-right on the MOV link that he provided and "save link as...". You won't have a problem. The reason why you think the download isn't working is because the file is 17GB, and doesnt include hinted streaming, so Firefox is trying to download the whole file before displaying it. Std web browsing here!
 
Josh Beadle said:
I would sign-on Pay for View for this if done right! Damn, my "girlfriend" costs me $3.95/min so this would be worth . . . like maybe $4.95/min!!

Thanks for the effort - you have the latest Alpha build on your Emachine?
Click to expand...

I think to make this relevant you need to keep sensor size out of it. Shooting 5k, 4k, 3k or 2k is totally different things it changes the DOF and actually precision of your lens and camera. When changing to a greater sensor size you most likly also introduce more DOF into the scene, the then out of focused areas are "easer" for the compressor to handle then the sharp areas. So then being on an open long lens 5K would benefit and get less noise. On a smaller sensor you would accommodate with a "zoom out" / wider lens to get the same frame with 2k... Also 2k is less sharp and demands higher precision in your glass...

So the "basic rule" is, according to me, to shoot as big sensor as possible and use as little compression as possible.... When saying that it is very rare that 5K 1:3 is the thing to choose... That is for harsh light conditions, high details and EFX shots and such. For all other scenarios where you are lit well and the only thing applied to the shot in post is grading and or look then you most likely are wasting disc space if shooting with 1:3. I find 8:1 to work fine for "normal" shots and then just bring on the extra data for the problem shots, like doing slow pans with a wide angle lens with a high f-stop straight into a forest with flickering leafs and you have a guy in the corner standing in front of a poor lit green screen... then you should consider nothing but 5k 3:1.

So I think it's hard to talk about the compression as a solo thing, it very much tangle up with lights, lenses camera moves, post and everything else.
 
It doesn't look like RED has any official "best practices" information on compression so let's keep this thread alive with updates as each of us is able to test on our own. With the cost of SSD memory and work flow logjam the terabytes can create - this is a topic I would like to see demystified and have science emerge.

Thank you for your efforts and continuing contribution!
 
Thanks Häakon, this is great. One takeaway is how surprisingly well 18:1 compression holds up. I think I'd like to see the same type of shot underexposed by a stop or two and see if I still feel that way once the images are corrected...

This is great! Keep it coming.
 
Jean Simenon said:
Just click-right on the MOV link that he provided and "save link as...". You won't have a problem. The reason why you think the download isn't working is because the file is 17GB, and doesnt include hinted streaming, so Firefox is trying to download the whole file before displaying it. Std web browsing here!
Click to expand...

Yes exactly, that is what you would expect. However still doesn't work despite a long wait as Sean Brown did. Anyway seems like it is an issue with my Internet, so nothing to do with this forum.
 
So anyone else out there want to do some more tests? Underexposure in particular would be interesting to see.

I'd post some except my Epic's sitting in the corner, awaiting a Canon mount.....

Haakon, any luck posting some R3D frames?
 
Very interested but cannot download the footage. None of the links are working. Help ?
 
links aren't working here either. Drobox is giving a 404 error if I try to load the video in a browser, and if I "save link as..." it downloads a tiny .mov file that clearly isn't the complete thing and won't play at all.
 
You must log in or register to reply here.
Back
Top