For best results use 8 bit depth per channel. If Maximum Bit Depth is used you may get by with 10 bits per channel. So that means 24 bit with RGB or 32 bit with RGBA if TIFF supports an alpha channel.
Final result will go to theatrical release,so I need higher bit depth.
Oddly enough if I convert the TIF file to a 16 bit PSD, Premiere will import it and will not complain about bit depth.
There is no way to keep 16 bit per channel on export, you have to go down to either 8 or 10 bits per channel, so why complicate matters? Were you in the Navy? That is where they made it SOP to complicate matters beyond recognition when there was a simple way to do things as well.
You need special equipment to present anything at more than 2 ^ 24 colors, which is standard for LCD, plasma, beamers etc. So 8 bit depth per channel.
And even if you find equipment capable of 67 mio. colors,it still is only 10 bit depth.
Do you realize that 16 bit depth per channel means 2.81E14 colors and nothing exists today to display that.
My TIFF sequences come from LR and AE. Both export sequences with 8 bitsperchannel or 16 bitsperchannel, never 10 bits. At the end I can open the Premiere sequence in an AE project set at 16 bits, so don't need to export from Premiere.
Well, it will be presented in a cinema, and before that it will be graded in a DI Lustre suite where I have to come with DPX files, or as an alternative with 16 bit TIFFs.
Don't know about the Navy, not from US.
The 2 standard for scanning 35mm to video, doing VFX work and exporting (doing a 'film out') digital video to 35mm at all the major film labs and visual effects houses I know of are DPX & Cineon. Premiere can import and export DPX (probably can import Cineon as well), so if you are worried about getting a theatrical quality, why don't you get a DPX image sequence from wherever you got the tiff sequence? If it's good enough for James Cameron, it should be good enough for you.
If you want to get egg-headed about it, here is a techy description of both:
It won't take a 16 bit .tiff - it will take a 16 bit .png
Do you have After Effects? If so, maybe you can get done what you need to in there.
Yes, at work I used to deal with dpx only, when exchanging scans from Kodak lab or other post houses but it was easy because I had Flame/Smoke and workflow was straightforward. Now I realised that LR doesn't export dpx files so I have to convert them to 16bit TIFFs, then load them in AE and export as DPX 10bit-log. Thinking that I have to do that for several hundreds of shots makes me sick.
The second confusing thing is that for some reason Premiere reports the DPX seq as having 24 bits depth, but I know for a fact that it's 10bit log.
What do you mean by that:
There is no way to keep 16 bit per channel on export, you have to go down to either 8 or 10 bits per channel, so why complicate matters?
Maybe do a comparisson between a 16bit TIFF>AE DPX>PRpro>DPX and 8bit TIFF>PrPro>DPX. zoom in all the way and look for jaggies on curved surfaces and color accuracy. You might be surprised, the difference might not be all that much.
No, the difference is big in terms of CC, especially going to cinema projection.
Now, the funny thing: I exported a 16 bit png file, import it in Premiere and in the properties panel it reports also 24 bit depth (instead of 48 as I would expected). So questions remain:
1. Why it doesn't import 16 bit tiff files, while in the help it states it does.
2. Why after importing 16 bit linear png or 10 bit log dpx, it reports them as 24bit depth (??).
3. If there is no switch to set the bit depth of the project, I must assume that it will try to go as high as possible, depending on the assets it has in the project?? This is confusing. In AE all these very important details are clearly resolved by switches and messages.
Sounds like an Adobe querstion. Is there any other other format you can export? I have been using prores for exports of AE effects to integrate back into my r3d timeline, works well. Color space holds up. but I am on a mac. I think QT Animation is pretty universal. talked to a colorist over at Warner Brothers who uses jpg images sequences, which suprised me.
Actually I've downloaded the Cineform Neo4k demo (Cineform does go up to 4:4:4 12bit linear) and was quite pleased with the workflow. But it's 999 USD, not the kind of money I would spend for a one time project.
Now, this link: http://blogs.adobe.com/VideoRoad/2010/06/understanding_color_processing.html clarifies a little bit things. It appears that Premiere just promotes everything to 32bit float, when needed, and applies various supported effects (the ones with 32 written on the icon). Now I hope that the author which says dpx 10-bit, has in it's mind 10 bit logarithmic.
Good to know about Neo 4K. I guess if I were a post house I would consider buying it, thankfully I don't think I need it.
After editing a short film and checking it for exporting to film print file... Here it is my experience:
1.- Premiere Pro CS5 imports TIFFs U16, so we could edit the whole stuff with 16 bits per channel in desktop mode (very slow for final footage but had proxies for animatic). Also accepts PSDs, but in this case we used PNG sequences.
2.- After several proofs for exporting i determine that CS5 works only in 8 bits mode !! So trying to export TIFFs or PNGs did not work in 16 bits.
3.- We use TIFFs in 8bits and fake DPX for a first print copy. Before checking footage we notice several pixelation because of the "bit downconvertion"
- Imported Premiere project into AE, without audio tracks.
- Before open the sequence, checked media to ensure that frame rate was the same used in Premiere. In my case had to change footage to 24fps (it is very fast).
- Checked all the video tracks, because some of them "displaced" -you will have one track per clip used in Premiere; some of them are SHORTER in AE timeline, only with IN point. OUTs are OK.
- Check transitions, as cross dissolves -NOT imported in AE-, to make manually with oppacity nodes. You will have to enlarge footage to apply nodes of transparecy.
- Render queue. Here you can choose several outputs at same time, rendering them in parallel. I chosed DPX U10, linear and logarithmic (one folder for each dpx sequence) and TIFF U16.
- Voilà. After 2 hours rendering 3 outputs (writing two dpx folders and the tiffs one at same time, with a total of 30.000 frames in 200gb) we had the same number of frames than exporting in Premiere, but in this case with real 10bits DPX files for transfer! And 16bits footage for keeping in case of emergency!
- Be sure to check the whole clip with a DPX player (will not work in real time but it is suitable for checking slowly).
Thanks for sharing....!!!
I am having the same problem working in a short film with many VFX... I need to export a lot of clips in "10bits Log DPX" from my RED Sequence in Premiere.... I tried with every parameter in DPX export, then reimport in my project and choose Properties: 24 bits pixel depth ¿...? Everytime....
So I am trying to open my Pr project in AE... but the only option to do it is by Dinamyc Lynk... is that correct? ...what I understood from Adobe is that you can open it as a clips sequence, but I don´t find the way... ...
ok... I found it!!! Now open as a sequence... change the project to 32bpc and export the clips I need for VFX as a DPX Over Range Logarithmic, no working space... I open this in Premiere (just to check the properties) and still says 24 bits... shouldn´t it be 32bpc? I tried changing the Gamma to RedLog in the original R3D but the same happens.... Could this be an issue in Premiere?
What am I doing wrong??
Hi K rrusel
There is no way to export 16 bits from Premiere... I tried a lot of things... I was lucky because the short was only 7 minutes.
Even if you export 16 bits (per channel) sequence from AE and try to import into premiere project it will be recognized as 8 bits (so 24 depth is shown).
I have checked other problems in CS5... Previous versions could strech audio and get good conversion changing percentage and retiming. For example, from 24fps to 25fps.
So solution now, for me, is using Adobe Soundbooth!
It is supposed that an editing soft as Premiere will offer only 8 bits per channel for broadcasting, although vfx and post have been made with higher depth.
In my case i needed to transfer to 35mm and a DCP. I did it with AE as i described and worked fine. The DCP is great! And made 5 prints with Cinevator Five system.
I believe that in the properties display that you are looking at "pixel depth" is referring to the
total pixel depth and not the per channel depth 8 bit per channel x 3 channels = 24
Hi again Salazaroo....
Thanx for your answer...
I am working in a project shot with RedOne and the output will be 35mm and DCP.... we´ll made the prints with Cinevator system as well... same as yours...
What I did is open the Pr project in After effects and from there export the R3d as DPX to work in the compositings... I am trying different ways to do this... change the R3D settings to RedLog and RedSpace before to export as DPX OverRange, or PDLog985 and RedSpace before to export as DPX Standard (as Mark Christiansen says in THIS SITE) ...
Once I do it, I import the DPX to start the composition... so I change the Cineon Settings to OverRange or Standard (depending of the settings at export) and apply the Cineon Converter LogToLin... but then everything looks too overbright to work... I am working with the DPX and .psd and tiffs together so it is like crazy to match everything together...
I tried to export as PDLog985, RedSpace and OverRange Cineon Settings (instead of Standard) and it looks much better... I check the picture looking at the Info panel and everything is Ok, or close the 'Iris simulator' and detail appear in the image, ...but when I check different elements coming from matte painters and 3d artist, whites clip at 1... how can i match everything together in the same space color to work and then apply the Cineon Converter LintoLog before export???
This is the first time I work with DPX and I am reading a lot about it, but I could´t find a step by step explanation... and it is like there are many ways to deal with it... some people says to use a general Working Space, some people says don´t (See this Article) ...
Thank you man.... I really appreciate your help....
Based on experience... I am sure that you can work in linear mode, till printing. I mean that log or lin parameters have sense when you are scanning from film, postprocessing (in linear, with LUT for it), and then transfering again to film... There is no difference with video material, or stuff from SLR camera with RED1...
So i think that if you keep depth higher than 8 bits, you will have enough range for transfering without problem.
If using DPX log as working mode you will have a very brighty image, because of the curve, that is designed for keeping range for/from film.
You can work with TIFF sequence, without compression. And use DPX u10 or TIFF u16 for transfer. But only for conforming. I am sure there will be other people working in different ways.
Imagine that you are transfering to film one picture retouched, from Photoshop, in 16bits. It is a linear picture. So, more or less WYSIWYG in that frame. Other thing is the screening with projector, that can be different; but a DCP, projected with digital projector will be the same you can watch in a callibrated monitor.
Sorry for my english! Hope to help u...