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 artifacts in slo-mo

K

Karl H

Well-known member
Joined
Jan 5, 2007
Messages
612
Reaction score
0
Points
0
Hi All

I've just received my stage-3 package and overall I'm pretty happy with it. I've only had 1 day of testing so far and we run kind of a 'stress test' shooting high contrast, high detailed shots, multiple frame rates and resolutions - just to see where the strengths and weaknesses of the camera lie. We want to know what will pose a problem and what will be fine.

Shooting at 120fps in 5K looked good in some situations, but also produced some bad compression artifacts, especially in detailed shadows on a high contrast shot. I know this is being crammed into 12:1, and the area is obviously part of the underexposed portion of the image (as shadows would be), but the detail in them seems to cause a big problem for the compression engine. It's not sensor noise. The odd thing is when the camera is static the compression is worse, but when the camera tilts up and is moving, the compression seems to cope better. Almost like movement is easier for the compression engine to handle. Even when we scaled this down to 2K output it was so distracting it made the shot unusable IMO. Has anyone else experienced this?

Shooting 2K 300fps was a different story altogether, it was completely unusable in the same scenario above with a high contrast detailed shot. The blockies are way too much and virtually the whole shot looks like a low bitrate mpg. However, we took the camera indoors later and shot some slo-mo against a clean dark background. This time the compression engine didn't find this a problem and it looks pretty good for 2K.

I just wondered whether anyone was following specific guidelines or rules for shooting slo-mo in 5K and below? To me (and this is a first test only) results can vary dramatically from pretty stunning looking, to muddy and blocky depending on the amount of detail you throw into the shadows. A dark blank wall indoors looked fine on playback, but the tarmac outside brought out the nasties. Here's a couple of images from our test, one full 5K jpg where the shadows are full oF artifacts and a 2K in which the whole image looks poor. (the blockies are not from the jpeg compression.)

http://www.darklineonline.co.uk/uploads/2k300test.jpg

http://www.darklineonline.co.uk/uploads/5k120test.jpg

EDIT : just realised you can go up to 7:1 on 2K 300fps and we shot this at 12:1 (as we had just exited 5K slow mo). So will retest that particular shot with lower compression...
 
Karl H said:
Hi All

I've just received my stage-3 package and overall I'm pretty happy with it. I've only had 1 day of testing so far and we run kind of a 'stress test' shooting high contrast, high detailed shots, multiple frame rates and resolutions - just to see where the strengths and weaknesses of the camera lie. We want to know what will pose a problem and what will be fine.

Shooting at 120fps in 5K looked good in some situations, but also produced some bad compression artifacts, especially in detailed shadows on a high contrast shot. I know this is being crammed into 12:1, and the area is obviously part of the underexposed portion of the image (as shadows would be), but the detail in them seems to cause a big problem for the compression engine. It's not sensor noise. The odd thing is when the camera is static the compression is worse, but when the camera tilts up and is moving, the compression seems to cope better. Almost like movement is easier for the compression engine to handle. Even when we scaled this down to 2K output it was so distracting it made the shot unusable IMO. Has anyone else experienced this?

Shooting 2K 300fps was a different story altogether, it was completely unusable in the same scenario above with a high contrast detailed shot. The blockies are way too much and virtually the whole shot looks like a low bitrate mpg. However, we took the camera indoors later and shot some slo-mo against a clean dark background. This time the compression engine didn't find this a problem and it looks pretty good for 2K.

I just wondered whether anyone was following specific guidelines or rules for shooting slo-mo in 5K and below? To me (and this is a first test only) results can vary dramatically from pretty stunning looking, to muddy and blocky depending on the amount of detail you throw into the shadows. A dark blank wall indoors looked fine on playback, but the tarmac outside brought out the nasties. Here's a couple of images from our test, one full 5K jpg where the shadows are full oF artifacts and a 2K in which the whole image looks poor. (the blockies are not from the jpeg compression.)

http://www.darklineonline.co.uk/uploads/2k300test.jpg

http://www.darklineonline.co.uk/uploads/5k120test.jpg

EDIT : just realised you can go up to 7:1 on 2K 300fps and we shot this at 12:1 (as we had just exited 5K slow mo). So will retest that particular shot with lower compression...
Click to expand...
299 FPS used to work with less compression than 300FPS, alwways try to use low compression 5:1 or less if at all possible.
 
I'm glad you brought this up. I've seen some of this as well at even 60fps.
 
Just curious...does epic not let you go higher than 12:1 on 5k 120? 12:1 has a tough time in high contrast scenes. I only shoot a scarlet, so I was curious.
 
Opps sorry, I meant lower. Can you shoot 8:1, 6:1, etc.?
 
I was just shooting some water falls and rapids up at Yosemite last week on my Epic and was getting crazy compression artifacts shooting at 2K 240fps 6:1 REDCode. All the 5K 96fps 12:1 REDCode footage looked pretty good. But some of the rapids at 240fps were junk. Granted there was a lot going on in the frame but the Epic should be able to handle that.
 
Thanks for the responses. I haven't done more testing yet, but from what Im hearing and my initial experience, it seems this isn't a foolproof slo-mo feature. Subject matter and lighting conditions will greatly affect the quality of the results.

Mark, 12:1 is the lowest it will go in 5K, lower compression ratios are not avilable. You can go higher to 18:1, but from my findings you wouldn't want to.

Perhaps Red would consider a non-RAW capture for high speeds? Would that help at all in keeping the image free from compression artifacts? Just thinking aloud. I would be very cautious to use it as is with a client present.
 
any solution is going to be costly...ie the new 512gb SSD cards. This will help compression ratios come down, but jarred hasnt announced by how much, and the price of these cards. I'd like to take this opportunity to say like the Bomb eyepiece these system fixes are dipping deep
into everyones pockets perhaps RED may show some love with these readdressed items that imho should have been part of the system before the entire Epic system was originally released...
 
Andrew Walker said:
I was just shooting some water falls and rapids up at Yosemite last week on my Epic and was getting crazy compression artifacts shooting at 2K 240fps 6:1 REDCode.
Click to expand...
Waterfalls are a huge torture test for any compression system, particularly JPEG2000. What happens at 3:1?

I agree with Sergio that the new 512GB SSDs will be a killer solution for intensive footage like this.
 
Karl H said:
Hi All

I've just received my stage-3 package and overall I'm pretty happy with it. I've only had 1 day of testing so far and we run kind of a 'stress test' shooting high contrast, high detailed shots, multiple frame rates and resolutions - just to see where the strengths and weaknesses of the camera lie. We want to know what will pose a problem and what will be fine.

Shooting at 120fps in 5K looked good in some situations, but also produced some bad compression artifacts, especially in detailed shadows on a high contrast shot. I know this is being crammed into 12:1, and the area is obviously part of the underexposed portion of the image (as shadows would be), but the detail in them seems to cause a big problem for the compression engine. It's not sensor noise. The odd thing is when the camera is static the compression is worse, but when the camera tilts up and is moving, the compression seems to cope better. Almost like movement is easier for the compression engine to handle. Even when we scaled this down to 2K output it was so distracting it made the shot unusable IMO. Has anyone else experienced this?

Shooting 2K 300fps was a different story altogether, it was completely unusable in the same scenario above with a high contrast detailed shot. The blockies are way too much and virtually the whole shot looks like a low bitrate mpg. However, we took the camera indoors later and shot some slo-mo against a clean dark background. This time the compression engine didn't find this a problem and it looks pretty good for 2K.

I just wondered whether anyone was following specific guidelines or rules for shooting slo-mo in 5K and below? To me (and this is a first test only) results can vary dramatically from pretty stunning looking, to muddy and blocky depending on the amount of detail you throw into the shadows. A dark blank wall indoors looked fine on playback, but the tarmac outside brought out the nasties. Here's a couple of images from our test, one full 5K jpg where the shadows are full oF artifacts and a 2K in which the whole image looks poor. (the blockies are not from the jpeg compression.)

http://www.darklineonline.co.uk/uploads/2k300test.jpg

http://www.darklineonline.co.uk/uploads/5k120test.jpg

EDIT : just realised you can go up to 7:1 on 2K 300fps and we shot this at 12:1 (as we had just exited 5K slow mo). So will retest that particular shot with lower compression...
Click to expand...




Absolutely the same observations here.
Can't use 2K 300 fps, it is very soft and blocky.

Also had some issues with 5K 120 fps, but one time it also had to do with the camera not being warmed up enough. It really has to be running at least 10 minutes to get up to the temperature the blackshading was done on, about 32 degrees Celsius usually at the moment in springtime europe.
 
But other than some issues here and there, i am pretty much in love with the RED look.
Just ordered a sister scarlet for 2nd unit :-)
 
Likely the compression algorithm used for high speed is not the same. It makes sense. Even zip has a more/slower option and a less/faster option.

AT 5K 120 that's a lot for the Asic to deal with in real time, much more than. say Red One Mx was dealing with at 4K 16:9 24p. When you think about it, it's amazing that it works at all, and that most of the time it looks fine.

The new turbo cards will certainly help, as what you would want is to compress LESS at high speed (because the Epic is likely needing to switch to another faster but less perfect algorithm when doing high frame rates) , not more.
 
You must log in or register to reply here.
Back
Top