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 :)

Why One Light?

J

Jim Brennan

Well-known member
Joined
Oct 28, 2007
Messages
79
Reaction score
0
Points
0
Website
www.hooligannationproductions.com
I'm getting ready to help edit a short shot at 4K. I've always taken the R3D file, done a one-light in RedCine and output prores 422 and done an online edit in FCP.

The director, who will be doing a lot of the post work asked if we need to do a one-light, or will we be better off taking the R3D file as-is and exporting it that way.

He wants as much room as possible in post to play with the footage, but we are a little restricted as he will have to be working on my system (He's a Vegas guy). SO we want the simplest workflow possible, (which Prores in FCP seems to be), but still be able to tweak the images.

Suggestions?
 
NO. The lower bitrate (10), and 4:2:2 color do not give you sufficient range to handle and manipulate all the data in the RAW file. If you do go this route, you must do some basic contrast control and color temperature balancing at the very least.
 
Always do a one-light before exporting, get the image somewhat close. As Mike said, basic contrast control and color temp are an absolute minimum. Once you export to ProRes or other 4:2:2 or lower codec, you're throwing away a LOT of information and "room" to work in.
 
I like that answer.
Thanks Jeff.
 
Good advice. But what if you aren't going ProRes. If you go full uncompressed to take them to a pro colorist, are you still better doing a one light or just leaving them pure?
 
jimhare said:
Good advice. But what if you aren't going ProRes. If you go full uncompressed to take them to a pro colorist, are you still better doing a one light or just leaving them pure?
Click to expand...

Not going ProRes... Doesn't really matter, you are doing a conversion, and in most cases will be throwing away information when you render out to another format. At this point, there is no way to leave it "pure". RED footage is RAW and all current export options apply a color / gamma curve in order to relate the RAW data to a viewing gamut via REC709, REDSPACE, etc.. Actually, that's the way RAW has to be viewed no matter what. The RAW luma and color data must be put into context within a set of controls to make it viewable. Eventually, hopefully after the SDK is made available, other NLEs and finishing applications will natively support REDCODE. This will eliminate the need for pre-processing footage -- applications can read the original file and access all the RAW data for color grading. Hence placing the RAW data into a viewable context within their application space in a way that's relevant to their grading tools. Of course, SCRATCH already has this advantage.

If you don't want to do a one-light in REDCINE or RED Alert, then you will want to export to a format that preserves as much of the original information as possible. For still image formats, you're looking at using TIFFs or possibly OpenEXR, if that works for you. DPX and Cineon are 10bit and you are losing a bit of headroom by exporting. TIFF is the only uncompressed output format that will preserve the full 12bits of color data at this time. DPX comes close, but is compressed. Its 10bit is logarithmic, but you're still losing info. You may be able to get desired results from an uncompressed QT, but I haven't investigated that option. All that said... IMO, the best output format from REDCINE / RED Alert is DPX after a one-light CC. In a Mac workflow, a one-light CC and then out to ProRes HQ is awesome for taking that right to a full edit and output for broadcast HD and DVD.

If you don't do a one-light CC, Taking the R3D files to a pro colorist is great, if you can do that and the colorist can work from them. Otherwise, uncompressed 16bpc RGB or better is great too. At this point, we're starting to get into a full workflow discussion though. But I'll just put it this way. If you're rendering out of REDCINE into clips that will be used for online and edit all the way through post, you're skipping a step and sacrificing a chance to improve your final product by not doing a one-light CC when you process that footage. ...Just sayin'

The level of quality you're after is for you to decide. What is inadequate for one man is overkill for another.
 
So, if you've done a log and transfer into FCP and want to finish with the prores files you're severely limiting what you can do in color correction?

It seems like I read that some people think log and transfer to prores and finishing with those files is fine for an HD finish. However, with the project we're working on now we did a log and transfer and it seems like we're getting noise in some of the prores files that shouldn't be there.

Would conforming (using crimson) the edit into Red Cine, doing a one light, and spitting out to prores again help with this?
 
since the director asked, then no prores.

make the director sample cc in two consecutive shots(diff shots), to have a feeling and get used to it.make sure that u down sample the image with the right interpolation.

then, like the previews posts, either tiff sequence or dpx.
 
Harky said:
So, if you've done a log and transfer into FCP and want to finish with the prores files you're severely limiting what you can do in color correction?
Click to expand...

Yes. L&T is essentially just using the QT proxies, just a different way to bring them into FCP. You have far less control this way than if you were to go through REDCINE to ProRes and do a tweak or two along the way. Eventually this will change (I assume). No L&T available for Build 16 yet, so you're basically seeing a straight dump from the R3D file to the proxy equivalent in REC709, if I understand the current L&T module properly.

]It seems like I read that some people think log and transfer to prores and finishing with those files is fine for an HD finish. However, with the project we're working on now we did a log and transfer and it seems like we're getting noise in some of the prores files that shouldn't be there.
Click to expand...

That approach is "fine", but it's hardly the best solution in terms of image quality. The question is, does this method give you the results you want? Looks to me like it doesn't.

Would conforming (using crimson) the edit into Red Cine, doing a one light, and spitting out to prores again help with this?
Click to expand...

Yes. Highly recommended. My workflows of choice right now are:

1> Pre-process, one-light in REDCINE and output clips to ProRes HQ. Edit and finish in FCS.

2> Rough offline with proxies or quick ProRes (or DVCPRO) you generate with minimal or no adjustments. Output XML and translate via Crimson. One light new timeline in REDCINE and output ProRes HQ for HD and DVD projects. Replace proxy clips with newly generated clips and finish your project.

3> Same as #2, but output DPX sequences and use GlueTools to package your DPX into more usable form. Finish via FCS w/ Nuke and/or Shake.

#3 is my preferred method when I need better quality than ProRes delivers or I will be compositing with CG elements, chromakeying, etc..
 
"we are a little restricted as he will have to be working on my system (He's a Vegas guy)"

then it all comes down to - the director wants as much room to play with CC as your system will allow ( and if Time & $$ are entered into the it then you work within the limitations) .... IMO if this project is going to be for Tv/video/HD then it's likely to end up in 8bit world so converting the R3D clips to 10bit 4:2:2 isn't as bad as it seems ...the ProRes clips do have some room for CC ... if the Director wants to push some clips really hard & you start seeing artifacts perhaps on those shots you can go back to the R3D clips and push them hard and then render out to proRes again or DPX/Tiffs ?

like your Director i too use Vegas ... i take my 4k 2:1 files and convert to 2k 2:1 cineform files ... then edit in Vegas ( which is 8bit) .. most CC is no problem - on shots that i really go to extremes i may go back to the R3D files and push it in Redcine and again render out to cineform ... in my final Vegas edit output i turn on the 32bit float processing ... i do NOT scale 2k to 1920 - i crop the 2k to 1920 so my final image is 1920x1024 inside the HD 1920x1080 frame = black bars at top & bottom ... i then project the final onto a 8ft screen ( from computer DVI) ... no complaints here ... just got a blu ray burner and hope to be burning by end of the week ...
 
Thanks Jeff! That clears up a lot.
 
donatello,

That sounds like a good workflow for Vegas folks. Since you use Vegas and are talking Blu-Ray, do you have the new DVD Architect Pro 5 software? I'm definitely interested in your thoughts on it if you do.
 
Right now the director is considering doing the one-light in RedCine, then exporting as a Motion JPEG A to work with that in Vegas for an online.

My first question is whether this is a good idea to begin with. My second question is whether anyone has any experience with that format because exporting at "maximum" you get what looks like a 1%-3% increase in noise (mostly in the blacks) over the "Lossless" export. But the "Lossless" is about 3 times the file size.
 
Hmm, Jeff, I'm curious about you saying to *always* do a one light. I just did a short where I mastered it in 2k. Since I wasnt pushing it too hard, Prores worked fine. I did a 1k offline then in Crimson used Redline to output 2k ProresHQ files all completely flat. I thought that by exporting them from Redline flat (as opposed to pulling them into Redcine, which I hate due to its consistent bugginess) would give me as much room as Prores has to offer.
 
" exporting as a Motion JPEG A to work with that in Vegas for an online."

are you exporting as 4k or 2k ?

"exporting at "maximum" you get what looks like a 1%-3% increase in noise (mostly in the blacks) over the "Lossless" export ...

seems a small increase of noise over the R3D raw file might be normal... AND are you seeing this while it's playing back 24fps or looking at a still frame ? ...
i do see more noise in the darks in Vegas preview window on LCD monitor vs Redcine on same LCD but i do not see it when i set vegas up to send out the preview to digital projector ( also note in Vegas you'll see better image when setting preview to "good ..Full" vs the default "preview ..auto "

over at OneRiver media they did some codec test early 2000 .. i did find this under motion Jpeg

"Both M-JPEG A & B seem to be some type of spatial codec because if you pause a frame, move or change the QuickTime window and go back, it will band. Leave it alone or scrub the time bar and everything looks great. Very strange. Now look at the x10 generation test... the tenth generation is virtually as good as the first! "

Note: that M- Jpeg is listed as a 4:2:2 compressed codec ...
while photo- Jpeg set to 100% quality is listed as 4:4:4 and when set to 75% is listed as 4:2:2 .....

i haven't tried any of the Jpegs (yet) ...will be trying them out after 23rd ..
i see Mcgregor listed his Vegas workflow (which i'll also give a try after 23rd)

http://www.reduser.net/forum/showpost.php?p=250617&postcount=6
 
one point to clarify, uncompressed sd is better than any 4:2:2 hd compression codec in terms of quality.
 
Exporting at 2048x1024.

The noise seems consistent to me both in a still and while playing. I tried a different clip with fewer shadows and blacks and there seemed to be no difference until you really look into the dark areas. We may have to do some one way and some the other depending on the amount of blacks in the shot.

And yes, apparently it provides only spatial and not temporal compression:
http://www.siggraph.org/education/materials/HyperGraph/video/codecs/MJPEG.html

Thanks for the tips so far. Any other input is welcome
 
You must log in or register to reply here.
Back
Top