I am shooting on a Red Epic Dragon 6K and have a pretty basic question about compression. Does shooting at the lowest compression, say 4:1 make a really significantly noticeable difference vs shooting at say 14:1? Since the end result would likely be H.264, I am wondering if it is worth using all the hard drive space. Also would 14:1 take up more computer resources to edit (heard this somewhere not sure if it's true)
Welcome to our community
Be a part of something great, join today!
Compression question. Does it make a difference? 4:1 vs 14:1
- Thread starter Lee M
- Start date
Mike P.
Well-known member
I've noticed that a higher compression ratio is actually easier for a computer to chew through.
There is a difference between 14:1 and 4:1... but will it be noticeable after 6k>4k downscale and h264/HEVC compression? Probably not... If it's anything creative, I'd use ~8:1 as a safe bet (with 5:1 being ideal as it's visually lossless). If it's corporate video/talking heads/nothing that's going to be pushed around too hard in post, I'd crank it to whatever I was comfortable with.
There was another thread recently that said when the compression ratio is over encumbered by a scene's detail (that is, it's too much information for the given compression ratio to properly capture) it actually manifests as slightly buzzed-looking focus, rather than noise/macroblocking (which is typical of non-raw codecs.)
There is a difference between 14:1 and 4:1... but will it be noticeable after 6k>4k downscale and h264/HEVC compression? Probably not... If it's anything creative, I'd use ~8:1 as a safe bet (with 5:1 being ideal as it's visually lossless). If it's corporate video/talking heads/nothing that's going to be pushed around too hard in post, I'd crank it to whatever I was comfortable with.
There was another thread recently that said when the compression ratio is over encumbered by a scene's detail (that is, it's too much information for the given compression ratio to properly capture) it actually manifests as slightly buzzed-looking focus, rather than noise/macroblocking (which is typical of non-raw codecs.)
Last edited:
Michael Tiemann
Well-known member
The only objective thing I can tell you is that 14:1 takes less computer resources to edit because there's less image to assemble from the data. As for the subjective (i.e. noticeable/objectionable degradation), that's incredibly scene and delivery-dependent. There are many on this list who default to 12:1. There are many others who default to 5:1. The actual default from a fresh install is 8:1. Some push down to 2:1 whenever they can. I did some of my own experiments and found that for my typical scenes, my typical delivery, my own aesthetic, 12:1 and beyond gives me more noise than I want to play with, so I tend to shoot 8:1 for a nice margin of error. (I don't have Nuke and don't really want to bring it into my workflow.) I'm not mad at 8:1 compression at all (albeit on my 8K Helium and Monstro sensors). Compression is definitely an aspect of your camera that you need to learn via direct experience, not received wisdom from the Internet.
Björn Benckert
Well-known member
If I'm not just playing around I always set it to full throttle. Thinking capture can never get too clean. And better be safe than sorry, sometimes difficult to predict when noise will kick in. And more so on dragon than Monstro Monstro is way more forgiving and can handle much higher compression rates as it's less noisy.
- Banned
- #5
Mike Nagel
Banned
Ryan Sims
Well-known member
For several years I shot at 12:1 on most projects that were going to the web. I only had 2 64Gb Redmags so I needed to budget my bits. That worked well most of the time. I would sometimes see macro-block compression artifacts in under-lit smooth areas like a flat wall in a house. When I got some more Redmags, I switched to 10:1 for general scenes and 8:1 outdoors for shots with trees, bushes, water or grass in sharp focus. If you shoot, say 12:1 or higher with a human face and a tree both in focus, the compression algorithm will spend all the data on the tree's leaves edge detail and the human face will turn to macro-block mush (this is true for nearly every compression method).
That said, 8:1 seems to be a good middle ground. Not massive data to shuttle on and off hardrives when editing or to archive. But enough data to handle most scenes well.
But like Björn said, the best image will always be at the lowest ratio possible (3:1 - 5:1 range). If you shooting high-end work, you want the best raw image to work with in post.
Another note: always go to a lower compression ratio for high-frame rate (cropped sensor) work. I learned this lesson the hard way.
That said, 8:1 seems to be a good middle ground. Not massive data to shuttle on and off hardrives when editing or to archive. But enough data to handle most scenes well.
But like Björn said, the best image will always be at the lowest ratio possible (3:1 - 5:1 range). If you shooting high-end work, you want the best raw image to work with in post.
Another note: always go to a lower compression ratio for high-frame rate (cropped sensor) work. I learned this lesson the hard way.
Rodrigo Violante
Well-known member
I always shoot as high as possible, I really do not see any disadvantage in doing so if you have 2 mags of at least 512 each.
AndreasOberg
Well-known member
- Joined
- Oct 30, 2011
- Messages
- 1,674
- Reaction score
- 31
- Points
- 48
- Location
- Leicestershire, United Kingdom
- Website
- www.obergwildlife.com
I have done a bunch of tests and. You can see a real difference between say 1:2, 1:5, 1:8 and 1:12.
At every step you will be loosing detail. Especially the noise begins to get destroyed at 1:8 where it was a more fine grained pattern at lower it becomes more digital blocky at 1:8.
It also depends how much post processing you do, I work a lot with noise reduction and sharpening so quality is better. I think I would avoid 1:14 unless you absolutely have to, try to stay closer to 1:10.
The truth is though that the best is really to do a few tests, show a typical scen with different compression then you do the post production you normally do and convert to H.264.
And yes, a higher compression will actually be easier on your system.
/Andreas
At every step you will be loosing detail. Especially the noise begins to get destroyed at 1:8 where it was a more fine grained pattern at lower it becomes more digital blocky at 1:8.
It also depends how much post processing you do, I work a lot with noise reduction and sharpening so quality is better. I think I would avoid 1:14 unless you absolutely have to, try to stay closer to 1:10.
The truth is though that the best is really to do a few tests, show a typical scen with different compression then you do the post production you normally do and convert to H.264.
And yes, a higher compression will actually be easier on your system.
/Andreas
AndreasOberg
Well-known member
- Joined
- Oct 30, 2011
- Messages
- 1,674
- Reaction score
- 31
- Points
- 48
- Location
- Leicestershire, United Kingdom
- Website
- www.obergwildlife.com
I have done a bunch of tests and. You can see a real difference between say 1:2, 1:5, 1:8 and 1:12.
At every step you will be loosing detail. Especially the noise begins to get destroyed at 1:8 where it was a more fine grained pattern at lower it becomes more digital blocky at 1:8.
It also depends how much post processing you do, I work a lot with noise reduction and sharpening so quality is better. I think I would avoid 1:14 unless you absolutely have to, try to stay closer to 1:10.
The truth is though that the best is really to do a few tests, show a typical scen with different compression then you do the post production you normally do and convert to H.264.
And yes, a higher compression will actually be easier on your system.
/Andreas
At every step you will be loosing detail. Especially the noise begins to get destroyed at 1:8 where it was a more fine grained pattern at lower it becomes more digital blocky at 1:8.
It also depends how much post processing you do, I work a lot with noise reduction and sharpening so quality is better. I think I would avoid 1:14 unless you absolutely have to, try to stay closer to 1:10.
The truth is though that the best is really to do a few tests, show a typical scen with different compression then you do the post production you normally do and convert to H.264.
And yes, a higher compression will actually be easier on your system.
/Andreas