Is it possible to use autofocus for an entire movie?

In my pre-occupation with other things in my life at the moment I’ve forgotten to publish the results here of trying to shoot an entire movie with autofocus.

Another movie with Dominic Brunt and he agreed to me using the Canon C300-2 with stills lenses for the entire film.

Although I had an AC he spent a lot of his time watching me try to focus pull using a Samsung Tablet!

I wanted to do it myself just to see how difficult it was and what the najor pitfalls were.

It was a hell of a learning experience but I’d do it again, maybe with a few small changes like taking my own router just for the camera WiFi, using frequency scanners more often and killing anyone who decided to use any form of wireless transmission without clearing it with the camera crew first!

The main lesson is that the newer a lens is the better it copes, well that’s a shock!

Seriously, in a lot of situations the lens just couldn’t focus fast enough, actors running at camera in very low light was a serious issue.

Anyway, lots more here… http://www.cinematography.net/Canon%20C300-2%20Digital%20Assisted%20Focus%20on%20a%20Movie%20%20.html

 

CML and image quality – chickenshit’s

CML was set up for high end pro’s and those who aspire to be in that class.
I’ve never worried about offending anyone else.

There are lots of places on the net for cinematographers who don’t care
about or maybe just don’t recognise a high quality image.
I never wanted CML to be one of those places.

My recent experiences of working with post people whose whole approach to life was “that’ll do” “good enough” “nobody”ll notice” has caused me to stop and take stock.

I feel that CML has been drifting in that direction.

I realise the political and economic pressures that are on us, believe me!

That doesn’t remove from the fact that if we don’t stand up for image
quality nobody will we are “the guardians of the image” and painful though it might be at times we have to fight that fight.

Answer for yourself a few simple questions, in a world where data size
didn’t matter, where RAW recorders were tiny and cheap, where transfer times were zero, where processing power was vast WOULD YOU EVER SHOOT ANYTHING OTHER THAN RAW?

Of course you wouldn’t because deep down inside you know that compression damages your images.

Next question, bearing in mind the conditions listed above, given a choice
of a system that recorded equal amounts of RGB and one that recorded 50%G and 25%R&B and then guessed what was in the holes that that approach left behind would you ever use anything other than the full RGB system?

Of course you wouldn’t because you know that resolution and colour are
compromised by CFA systems.

Now, given that you would go the quality route every time, why are you being such chickenshits and compromising your images all the time?

Devastated 3

So, I’ve finally got a version of The Taking that I’m happy with, I certainly didn’t think that this was going to be possible after the C&C showing in Leeds on the 5th November!

It’s taken a lot of work to clean up the mess and in the process I’ve learned a huge amount.

I’ve also had a lot of help and I’d particularly like to than  Nick Shaw and Gavin Greenwalt for the LUT’s that they created to fix some of the initial errors. These errors were introduced when 4K TIFF’s were created for the VFX guys from the original R3D files, why the match graded log DPX files I had created weren’t used is a mystery. Anyway, the TIFF’s were created with the contrast, saturation and FLUT settings burned in that we had used for monitoring only on location and that were never meant to be used anywhere else.

This created a wonderful set of problems as there was more than one look used on location…

These files had been used by the post house in the final finish for the C&C showing, it should have been clear to a blind man that the files were not “normal” log files but whatever…

I’ve spent 4 weeks going through the edited DPX files matching black levels and contrast, replacing shots that had been clipped form data to TV levels and generally patching things up.

Finally on New Years Eve I had a version of the film that looked as it should.

A version that didn’t have to apologise for being low budget but that looks good regardless of budget.

It looks the way that Dom and I agreed it should back in July when we graded it at UWE.

 

 

Devasted 2

Wow!
Well that’s got to be the fastest and biggest response I’ve had to a post.
So let me clarify…
People don’t walk out of a showing saying that post screwed up the skin-tones, they say what a useless DP.
People don’t walk out of a showing saying that the editor zoomed too far into an image, they say that the AC couldn’t get it sharp.
People don’t walk out of a showing saying what a dreadful DCP, they say that the colourist screwed up.
People don’t walk out of a showing… I think you get the point now.
It’s not the people responsible that get blamed, it’s the image originators.
Well I’m not taking that.
Is my last post abusive and offensive? I hope so!
If we, the cinematographers, don’t stand up and scream when our images are messed up them nobody will and the rush into mediocrity will accelerate.
We are “The Guardians of the Image” and if in general we’re too chickenshit to shout when our work is damaged then we deserve what we get.
To all camera crew out there, stand up and be counted!!

Devastated

I’ve delayed writing this since I was at the cast and crew showing of The Taking.

I needed time to calm down and stop screaming.

The incredibly hard work of all the camera, lighting and grip crew has been devastated by incompetent uncaring post.

The DCP had obviously been made by a blind man who had left both his white stick and seeing eye dog behind.

Over saturated, weird gamma, blacks crushed, skin looking dreadful, thanks guys you just made our work look crap.

I’ve seen better looking grannies TV’s!!!

I graded the film and delivered 10 bit 2K DPX files, 2 versions, one log without film convert applied and one finished totally. I’d wanted to deliver higher res files but that was vetoed by the post supervisor.

During the conform and grade I’d reduced some of the extreme reframes that the editor had done, trying to keep the directors intentions intact but also trying to preserve image quality.

I didn’t agree with the reframes but ultimately that’s not down to me, I felt they reeked of low budget TV soaps.

The files I delivered had been graded on a system that was calibrated with Lightspace CMS, £3,000 of calibration kit.

Apart from the ridiculous looking images in the DCP there were also all kinds of reframes, zooms in and out and it looked like they had been made from the DPX’s not from the R3D’s so the image quality was horribly degraded.

I have no idea of the route that the images followed once they left me, they should have had the VFX added to DPX files at 2K and the DCP rendered from them.

I strongly suspect that they went through a compressed intermediate format and that was probably HD!

I had added a degree of mid-tone sharpness in resolve and was nervous that I’d overdone it, I needn’t have worried, the post guys managed to soften off the pictures wonderfully.

It was 6 months under 40 years ago that I first worked directly for a TV station, I thought that people weren’t attentive enough to what they were doing but didn’t know beter.

It was about 30 years ago that I moved into high budget commercials and met a lot of resistance from camera crews because I came from docco’s and TV. At the time I thought that this was unfair, looking back I realise what they were worried about.

TV is an area of work where “good enough” and “that’ll do” reign, almost nobody cares about getting things RIGHT!

The old line of “why is TV called a medium? because it’s never well done” was reinforced by your approach.

I was asked during the Q&A before the showing by a show organiser who had seen a video copy of the film, why did it look so different from a TV series that had been shot at roughly the same time in roughly the same locations? why did ours look cinematic and theirs look…

I pointed to the 60′ screen behind me and said “because I shot for that not for a tiny TV” I had repeatedly during the shoot reminded people that we were shooting a movie not a TV show.

I’ve offered to spend my own time and money making a decent DCP and deliverables if they’ll send me the final DPX’s and audio files.

The Taking – The Grade…

Well, I’ve started the grade, first at home and later at UWE.
An initial problem, that I was aware of as we shot, is that we used in camera looks and they don’t translate to Resolve or any “real” grading system.
Both RED and ARRI have used their own format of “look” files in the past rather than using an industry standard one, ARRI have fortunately learned and their latest camera accepts industry standard LUT’s.
The problem that non-standard looks cause is that everyone has got used to the look both on set and in the offline, now we have to try and get something like that in the final grade but it’s not easy.
Sure we can debayer into Resolve using the look stored in the metadata but this is a destructive process and if we do that any detail that is in the RAW file in the highlights or shadows is lost.
Well as the whole point of doing a grade is to refine the image we can’t go that route!
Graeme at RED kindly sent me a LUT that can be loaded into Resolve that will created RED Gamma 3, this is great as I always have the underlying log image to pull data back from but I now have to try and re-create the contrast and saturation alterations that we added to the looks.
This is much harder than it should be, I’m guessing that pivot points are at different places for RED and Resolve contrast and that 50% saturation means different things to them both…
This has led me to start a grade from scratch and take a different route, of course this causes problems because everyone has got used to the look and now doesn’t want to see a change.
Oh the joys or RAW and log!

NAB 2014

It’s really sad that the obsession with numbers that was started as a marketing tool by one camera company has swept over the entire camera market.

Now, all anyone wants to know is has it got a big number, not what are the pictures like.

I guess that it’s easier for non camera oriented people to grasp the idea that a big number is better but it isn’t!

Just look at the NHK 8K demos, very sharp but no dynamic range and garish colour, it was like television from 40 years ago.

We have to get back to looking at pictures and not worrying about numbers.

This is why Dolby Vision is so important, it’s about better pictures and that’s all.

Testing Cameras

You can’t win here.
It doesn’t matter how independent you make your test, how many independent people are present someone will always say that you’ve rigged it or cheated.
The test I published in 2002 comparing Fuji, Kodak F900 and Viper were criticised by Kodak because their images were soft, yeah and they were the ones who scanned them soft!
It didn’t matter anyway as the tests were about Dynamic Range not resolution.
As I test more and more and more I know what to expect and I’m rarely surprised anymore.

http://www.cinematography.net/edited-pages/camera-match.html

My latest tests, these are half of a 800/3200 comparison of these cameras and are in that page to show differences between the colour of the cameras.

There are a lot of tests available in CML and in most cases RAW files are there as well.

Accurate pictures or not?

I see and hear comments all the time about accurate colour and gamma reproduction and I always find myself wondering why.

My job isn’t to make accurate pictures, its to make pictures that effect an audience.

They may be pretty, they may be brutal but the last thing they will be is accurate!

The pictures will reflect the story I am trying to tell or the product that I am trying to sell, they will have nothing to do with reality or accuracy.

I don’t want an accurate camera, just as I never wanted an accurate film stock, I want a camera that is malleable, that can produce images that can be changes in any way that I like.

I’ve had problems in the past because the curves I use to de-log images aren’t accurate, well of course they’re not! I never intended them to be.

Single sensor have problems due to the nature of the sensor cameras

These all, I repeat all have problems with colour and aliasing and moire due to the nature of the sensor.

There are ways to reduce these problems but nobody has yet avoided them all.

The sensor design is fundamentally at fault, it’s a pattern of RGB sensors with holes between the different colours.

These holes cause colour errors, someone has to guess what colour was in the hole, yeah I know they say they use sophisticated algorithms but that’s just a smart way of saying guesses.

The holes cause aliasing and moire which the manufacturers reduce with OLPF filters, optical low pass sound a lot better than diffusion doesn’t it?

Oh yeah, that’s what they’re doing, reducing the resolution to reduce the errors. Brilliant!

Just what I want, a camera with faulty colour and a diffused image…