ProRes 444/422

[Trench117]

Hi,

I am currently working with a designer to decide our workflow for an upcoming production. We will be shooting the content raw on location using a RED Epic.

The designer has asked how our server will handle ProRes HQ 444 files. He is considering lowering this to 422 and not shooting raw if our system is not going to 'handle' it.

The outputs will be three panasonics at 1920x1200 on an x4. At the moment they will be treated as three different mixes/outputs.

My current thoughts are to capture at the highest resolution possible then downscale if needed as this will still look better than being captured at the native size. This means we have future proofed the production and allows us to use all three outputs as one canvas (5760x1200) if needed although this seems to be pushing it performance wise.

I wondered if anyone had any experience with the use of 444 vs 422 and how it handles/looks in Catalyst.

Any thoughts or comments appreciated!

[Peter]

we're doing a lot of prores 422 LT or proxy lately
looks very good and plays well on 5.1 servers
we've already used a similar setup: panasonics DZ110 and x4
below quote from hugh

ProRes uses hyper threading which means it performs very nicely on modern Mac Pros. Don't bother with HQ. Nice for Post Production. Not much use for Catalyst as none of our displays/ projectors/ LED walls can render 10 bit.
Hugh

Peter

[Mr_P]

Catalyst will playback ProRes 444 - but is s rarely necessary...

The data-rate is the key consideration - you need to test to see what the maximum data-rate your system and SSD will playback... Not a lot of point choosing PR-444 and high resolution if your hardware cant play it back.

ProRes-422 is probably sufficient and will result in lower data-rates.
As Peter mentioned, ProRes LT or Proxy is usually ok

...and before anyone says it: '...well it plays back ok in quicktime'

Quicktime Player prioritise audio - to prevent glitching.
Catalyst prioritises video - different to Quicktime Player - to provide best playback'

The outputs will be three panasonics at 1920x1200 on an x4. At the moment they will be treated as three different mixes/outputs.

Confused - are you using 3 individual screens at 1920x1200 with unique content? or one canvas at 5760x1200?

If you have 3 independent separate surfaces - then your max resolution only needs to be 1920x1200 as you stated
- or are the 3 surfaces being used as one canvas?

My current thoughts are to capture at the highest resolution possible then downscale if needed as this will still look better than being captured at the native size.

Working with RED files can significantly increase your workflow and render time.

If you get the results you need at a lower resolution / quality - it will probably speed up your workflow.

No point capturing files larger than your projection resolution or canvas size...

This means we have future proofed the production and allows us to use all three outputs as one canvas (5760x1200) if needed although this seems to be pushing it performance wise.

Not really an issue of 'future-proof' - more a question of your show requirements...


S

[Trench117]

We have set aside some time in rehearsals to test 444/422 HQ,LT and proxy.

At the moment it is three screens with unique content on each. So 1920x1200.

The designer is keen to work in RED raw format and then downsize on the final export to 1080. It may well turn out that with his hardware setup this is not possible. We shall see.

In terms of capture size vs output size I have heard that downsizing from a larger resolution still looks better than capturing at the final output size? I have yet to see this for myself....

Also there have been questions asked about what bit depth catalyst can handle/process? Im not really sure how and where this comes into it. Does the projector not have the final bottleneck in bit depth?

[Spam Butterfly]

Redcode is as wavelet based codec that encodes the raw Bayer sensor information from the camera. It is not an RGB codec. Nor is it visually lossless. I guess whether you use this natively as part of your workflow is up to you, and how much grunt you have available to you. When Reds capture there is always a tradeoff between resolution, frame rate and compression ratio. Ultimately we aren't in Hollywood or Soho post-productions making feature films or high end adverts. So ProRes of some description is good enough for most normal mortals, particularly in view of the fact that most Macs can throw it around with considerable ease, and plenty of feature films, TV programmes and adverts have been made with it.

The Gorillaz (RIP) and the Roger Waters Wall show used ProRes in their shows. Two meticulous artists with respect to content you will not meet.

ProRes is supposed to be visually lossless and it there's a certain amount of industry consensus that it's pretty darn good. ProRes 4444 is useful for post production as is ProRes HQ. What you use after that is a moot point. ProRes 422 Proxy is better than Apple Intermediate Codec that we were all using 6 or 7 years ago.

As for colour bit depth - is your designer editing footage on calibrated monitors that can actually display 10 bit? Broadcast Grade 1 monitors working at Rec 709 use a restricted gamut BTW. If you are not grading/ editing and finishing your footage within a 10 bit workflow using 10 bit calibrated monitors then what are you worrying about?

Mac OS, Open GL and decent DisplayPort graphics cards support 10 bit if you have a monitor good enough to display it.

Given that most people have to go through DVI as an output format from the graphics card, it is worth bearing mind that DVI can only display 8 bit...

Hugh

[5th Element]

Not that any two projects are alike, but I often find myself asking in these situations, what is the end game? Where is this footage actually going to be viewed and then based on that I decide how important certain quality issues are. In most situations for me the biggest concern is staying in sync at the lowest cost. If I have lots of monitors playing back content spread over them I want it to be in sync way more than I am concerned with how high quality the codec is. Viewing distance and screen size are always in my mind when I consider these types of things. I may not be a video-phile but I doubt I could tell the difference between a 422 HQ and 422 proxy on a 50' plasma at 20' maybe it's just me.
But I feel like I am missing something, if you are shooting in red epic, which could be 5120 x 2700 and you want to have 3 x 1080 you would actually be scaling up to fit it horizontally while at the same time losing almost half your resolution vertically. Not all bad since your pixels will be close to 1:1 in the area you want. If it were me, I would shoot at the highest quality possible, edit at the highest quality possible, render out a full size 444 and then from that you can make lower quality slices if need be but you always have a high quality source file to work from on other projects. It doesn't take that long to do a re-render at a lower quality but you will regret it if you do the whole thing in proxy from the get go and later want to stretch the whole thing over 6 monitors instead of 3 one day.... but then again that depends on whether or not you think you might try that.

[Trench117]

At the moment it is going to be 3 separate outputs each at 1080. But like you said easier (workflow allowing) to start high and render out lower if needed. The issue with sync is a good point to bear in mind thank you!