On Saturday I saw the results projected and not just projected but on an IMAX screen! This film was never intended to be seen this size, we knew we were shooting for TV screens. I was sitting in the front row, not by choice but I had to be there so that I could quickly take part in the Q&A afterwards.
So, did it work? Well yes, no more soft shots than any other movie I’ve seen on the large screen recently. In other words not a lot at all.
The entire movie was shot either with autofocus and face recognition or controlled via a tablet with simple finger presses. We learned how to quickly switch between modes while we were shooting and also what we could do during a shot. The one thing we thought was missing that would have helped a lot was a couple of preset positions that we could load as escape positions. Preset so that if the face recognition lost track during a fast move towards camera we had emergency goto’s in the most important positions. Happily Canon have now added this facility.
I said at the time that I probably wouldn’t do a movie like this again, having looked at the results on an IMAX screen I now think that this is the future. At present you are limited in your choice of lenses and camera. Canon are moving in this direction faster than others. I can see how they could, if they wanted, interface the /i info from say Cookes to the cameras internal software and an external focus motor and so enable this for any lenses with /i.
As we move to bigger and bigger sensors we need to look very closely at this kind of technology. There are other solutions out there but they are all expensive and somewhat complicated. This was very very simple, even a cinematographer could operate it!
I’m finding more and more that I have to wait until night to be able to even try and watch some series. They are shot so dark that unless the ambient light is lower that it would be in a grading suite the whole thing turns into a radio show.
I think that there are a couple of issues at play here, first people, and that’s DP’s as well as directors, not understanding that an image doesn’t actually have to be dark to convey a feeling of darkness. In fact judicious use of bright area makes other parts look darker. The other issue is highlights that are too bright, or areas of a shot that are too bright and that not only make dark areas look darker but also draw your attention away from the main interest of the scene.
I think, and I’m stressing that this is a personal opinion, that there is a huge amount of outright incompetence out there at the moment. People who really don’t understand how to use images to tell a story. People who should not be in charge of a camera.
Yes I do have a home with white walls and light grey floors and white seating and lots of windows but even at night with curtains drawn the image is too dark to see WTF is happening on screen. Once we get HDR I’m going to have to paint the walls black change the floor and the furnishings, black the windows out totally and wear a burka to make sure that no light kicks back off my white face!
Guys, use a domestic set to check your grading and turn the f’ing lights up when you do, go out and watch it on a TV not a monitor in reception.
Ah, that feels better!
Oh and finally, I do have all the kit to set my screens up properly but I find more and more that I can’t use the “correct” setting on my home TV, I have to use a mode like Vivid! to see anything.
I keep getting asked what my conclusions are after these evaluations. Well, I uploaded the EXR’s so that you could make your own decisions. However, I do have preferences. These are the high end cameras from each manufacturer, there are 2 from Sony because they seem confused 🙂 These were from the original RAW files converted to EXR ACES linear AP0 in the manufacturers own software. They were then loaded into Prelight in a Rec-709 BT-1886 calibrated environment and the exposure that was closest to “correct” was chosen. As you can see the cameras exposures vary a little from the manufacturers recommendations. I then adjusted the colour to give a neutral result on the vector-scope.
No alterations were made to exposure, contrast or saturation.
The chosen frames were then exported in sRGB jpeg form. They are HD res.
So, which is best? I prefer the colour from the Alexa and the F65, I think the Helium changes colour most with exposure changes. The thing is that they’re not major differences and far more likely to have an impact is ease of use. In this case I prefer the Alexa, Varicam Pure and C-700.
The biggest shocks are in the lower cost cameras, they’re bloody good! I’m doing a long term review of the BMD Mini-Pro at the moment and am waiting for the Panasonic EVA before doing a comparison of the lower cost cameras. I have already rejected one camera in tis category because I hated the way it felt and it’s lack of ease of use.
I’m a member of a number of colour grading groups on the net and they vary from the great, LGG, to the unbelievably bad. I made a comment a long long time ago that the way to tell who was a dailies colorist was to look for the white stick and the dog. It seems that now everyone is a colorist and most of the so-called colorists are either color blind, partially sighted or completely blind. I see examples posted online asking for comments, I have refrained so far. My only comment would be “go and find an optician!” Guys, you need to start with a good image. If after loading your work into your software of choice ( usually Resolve) set the workflow to ACES and use the relevant IDT & ODT. If after this your work doesn’t look pretty good you have fucked up the shoot! go and start again. Great grading has to start with material the colourist can work with, yes a really good colourist can rescue crap work but there are limits. If you give them good work to start with… Please notice I am saying that you should give your work to a good colorist, not that you should do it yourself. If, for whatever reason, you end up doing it yourself then shoot it right and don’t do a lot in grading.
It seems that everyone using grading systems for the first time finds the 11 setting before anything else!
I am so fed up with seeing oversaturated images with crap skin tones and weird contrast. Learn to do it right first. Don’t use the “I am an artist” excuse to cover up bad work.
I’d recommend that you view these on your smartphone, it’s only 10.5GB in all and each file is at least 50MB but hey! your smartphone will easily cope with that!
If you do that you’ll see that there is absolutely no reason to get one of these cameras, your DSLR is way better and your iPhone better still.
Actually, I’m thinking of doing a similar test with some DSLR’s, I’d rather eat my own leg but I feel I have to do it just to prove a point.
On the other hand, maybe I’m wrong, maybe a $2K camera is really as good as a $60K one. If you believe that I’ve got this bridge you may want to buy…
Oh and just in case… load these files into Resolve with it set to ACEScc and 4K DCI, then set the input to ACEScc and the output to 709, don’t crop the image, move around it using the positioning controls in color.
This will let you look at the full beauty, or horror, of each file.
Yes, they all pretty much capture it all, well, a couple of horrors at the top end but you’ll find those for yourself. Now, look at the lower exposures and ask yourself how much noise and colour shift you can live with. You now have the USABLE dynamic range. Strangely enough it’s not always what it appears to be or what manufacturers say it is.
Well, I’ve uploaded QT files of all the cameras in HD and all the ones shot RAW in UHD as well.
They can be downloaded in their full unaltered by Vimeo form as well.
So how are people viewing them?
So far 606 total idiots have viewed them on their smartphones!
Unsurprisingly the Ursa Mini-Pro and the Canon C200 are the most popular however they also have the highest number of not completed viewings.
People viewing the evaluations via CML watch the videos to the end and tend to watch them at higher resolutions and download the original files far more than people going in directly to Vimeo or via other sites.
Why start viewing a camera evaluation if you’re going to give up after the first few exposures? one camera I can understand, you didn’t realise what was involved but the same person doing this on camera after camera? what kind of person does this, certainly not anyone who knows anything about cinematography. Anyway, WTF are you doing viewing them on a smartphone?
I will still go ahead and upload some EXR’s for the people who understand what their doing, precious few though it is!
We are about to start a new round of camera evaluations using a new standardised system that will allow future tests of cameras in identical conditions.
The cameras included this time are:-
Arri Alexa Classic recording to both internal QT and external raw
Arri Alex Mini
Arri Alexa SXT recording to Codex
BMD Ursa Min Pro
Canon C300-2 recording to both XF-AVC and external raw
Canon C700 recording to both XF-AVC and Codex
Panasonic Varicam LT recording both internally and external raw
Panasonic Pure recording to Codex
RED Weapon Helium
Sony FS7 recording internally and external raw
We had hoped to include the Panavision Millennium and Alexa 65 but the first is in the process of being updated and the second is just too busy in the limited quantities that are available. Both are scheduled to be included in further evaluations later this year.
The Nokia OXO will also be included partly in the “normal” tests and partly to live stream the tests for those who have nothing better to do than watch live streams of camera tests!!
Lights are from BB&S and the lens being used for all except the C200 is the Fuji 4.7 * 18 T2. We will be marking up the in-shot idents with the focal length used to produce an image size that matches “S35” with 50mm as the base focal length.
With a schedule of 4 cameras per day we will be much more relaxed than in previous evaluations and paying a lot more attention to detail.
All camera settings will be published with the results We will also transcode all the rushes into 16 bit EXR files in ACES using the manufacturers software wherever possible. If there is no manufacturers software we will use Resolve. I will also be using Prelight onset to grab frames from the monitoring output of the cameras as part of a very different test 🙂
Results will be published towards the end of July.
I must be in an alternative universe, I just can’t believe what I see and hear.
11 years ago I shot a movie digitally, it was shot in HD in true RGB and uncompressed. Each of the 3 sensors had 1920 * 4320 photo-sites that created 3 * 1920 * 1080 images. It was 10 bit log and 4:4:4 generating 447.45 GB per hour.
Now Netflix are specifying that they will only accept “4K” cameras but they’re not 4K, they’re 2K * 2K in green and 2K * 1K in R&B they’re barely 3K and that gives you with Redcode 5:1 a grand total of 189.10GB per hour. It’s not a problem of Redcode, it’s just it was easy for me to get those figures, all the cameras they accept produce similar data rates.
Now I accept that technology moves on and compression get better but 4 times the resolution and less than half the data! no chance.
The issue is that people making decisions don’t understand what they’re deciding and are swayed by marketing.