fwd:re: [tig] 4K for DI - Unappealing image quality?
Wed Apr 14 11:43:00 BST 2004
Don't worry about upsetting me, learning only comes with the sharing of ideas and views, and often that means disagreement. Also, my comments are meant to provoke discussion as I don't always practice what I preach. It's too easy to make a statement and never back down. I'm often proven wrong and am happy when that happens as it means I've learnt more of benefit.
I should also say the views in any of my mails are the views of Digital Praxis, based on a lot of hands-on operation at many DI locations around the world, and no one else. I also have to say that although I get to play with a lot of different equipment I have not used it all by any means. I'm still waiting to play with the new ArriScan for example.
And congratulations on the two iQ's. I would love to see your workflow as the Quantel boys tell me you have it well sorted.
I have re-read my message, and as usual, I seem to have not presented myself well... Comes from thinking about the evaluation in hand too much and not about the text of the message - eats shoots and leaves, you might say. I also meant to have a question mark after the subjet title - now added.
This evaluation was a view of the images being projected digitally, with no film projection. This was the basis for the final comment of 2K being better than equavelent 'film' projection as seen in normal film theatres where duplication occures - 1.2K vs. 2K, so digital is already beyond 'normal' film. I don't think I made that very clear.
Also, the evaluation was based on visual 'feelings' of those involved - very personal - backed up with vector scope and waveform monitor. This assessment was started by one of my clients, who I'm not sure want's to be named in fear of being tainted by association ;-) but it was their comments that prompted the wider assessment, with other users.
I didn't mention any equipment in use for the DI as this was not important, just the images on the screen. It could just as easily have been scanner direct to projection.
As this was digital projection the losses involved with film duplication were removed, and this is the crux of the comment. In the old Domino VFX days the final film output was cut back with the remaining OCN so the 'digital' film went down the Full duplication chain too. As a lot of the DI projects I work on have limited distribution they often go through a much reduced dupe process, so maintaining MTF values.
Additionally (and I didn't tell you this!), a lot of Domino work was done 1.5K due to time constraints with projects, and very few people ever spotted the difference in the final film.
As for 8bit YUV with dynamic rounding, that's a whole different issue to resolution, and shows what is possible with a well calibrated closed loop system running a limited, but understood, dynamic range. Men From Mars, a re-birth of my old Men In White Coats VFX post house in the UK, still use their Dominos for a lot of their film work, although they are getting a bit long in the tooth now!
I do think we will see 4K gaining in popularity, but I'm not convinced for the right reasons. I refer back to the 1.2K of an average release film print. Just because something can be done doesn't mean it has to be. I have a similar view with regard to the DCI's views on 12 bit 2.6 gamma projection vs. 10bit log with 'print' luts. I would rather see 10bit log with dynamic rounding and in-projector luts.
Mind you, 3K resolution does have appeal, especially if Nyquist down sampled from 6K scans.
Like Marc Wielage I have found 4K scans Nyquist downconvertedato 2K provide the best subjective images - and Pete, that means 2K at least back to film. Most film recorders upres. to 4K prior to film recording.
I also agree with Mark that grain reduction is a good tool, but I don't really want to use it on a full feature, and this is one reason why I'm keen on digital cameras such as the Viper, and with MotionFX in the UK am working hard in building a fully digital workflow for features, specifically lower budget UK ones where the benefits can be great.
And while print stocks have a slightly different perf. pitch, what of dupe to dupe? And that still doesn't deal with grains never aligning, which is probably the bigger issue in resolution loss and 'noise' reduction.
And while I agree grain is technically different to video noise, it is still seperate to the image, even though it is integral to its makeup. Just look at the 'grain noise' on a flat image, such as a test chart. If the grain wasn't there the image would be a lot less 'noisy'.
And I'd have to ask, how many clients ever ask for a 'data' archive of either the original scans or the data final? In all the DI projects I have been involved in the past 3 years (over 20) no client has ever asked, being ony interesed in the final deliverables. Sad, but true.
And when it comes to Spirit data scanning it's a very hard beast to tame. Thomson have never made calibration for data easy (as has Cintel) and the range of possibilities for quality is enormous. This makes Spirit 2K vs. any scanner a difficult comparison at best - and I've seen a wide rang in the past few years.
However, as with most things in life, there is no 100% right or wrong, just alternative views that work for the situation in hand. I and my clients are presently very happy with the business model, workflow and quality of a 2K DI operation, but that view may change as the market and the equipment servicing it changes. I already espouse the benefits of 4K to 2K Nyquist scanning, and think 6K to 3K would be very interesting to review!
However, I reserve the right to change my mind - again ;-)
It's also fair to say I'm a film romantic. I've just watched Under a Tuscan Sun on a crappy airline monitor and loved it. A good story can over come poor quality. Great quality can never over come a poor story - still working on getting films direct delivered to my palm!
And as a final comment, the breadth of knowledge on the TIG is very variable and it's difficult to pitch any comment to appeal to all. I don't always intend to offend - it just seems that way (insert yet another smiley here!).
All the best,
Digital Praxis Ltd.
+44(0)7765 400 908
steve at digitalpraxis.net
More information about the Tig