Thread: KOMODO....
|
|
08-22-2020, 02:56 PM
My dream is a sine wave electronic (typically referred to as global) shutter. Bye bye time aliasing, judder, etc. If DSMC3 is capable of giving us control over the shutter characteristic, think motion mount or Tessive Time Filter without moving parts, that would be the holy grail.Originally Posted by Phil Holland
Cheers - #19
08-22-2020, 03:50 PM
Big agree. I'm Dolby Vision Certified/Licensed after several years of rather painful bleeding edge HDR work. Things are settled a bit more now and I'm hoping we'll see more HDR in the future. Streaming is exploring it a lot. But after my last couple of workshops in 2019, it still seems the majority of the market hasn't worked in HDR just yet. More people will be doing it soon though.
That would be nice. There's lots of sensor tech out there as well as other hardware that can come into the mix in the future. At the core I think we as shooters don't want something built in that limits shutter speeds in any capacity by default, but who knows. There could be other ways.
I will say the Soft Shutter mode on the Motion Mount was/is useful in combating judder and an interesting take on motion blur. Temporal aliasing is the thing there.
I'm still doing array work and Komodo's global shutter represents something alarmingly useful in particular when dealing with larger arrays with lots of cameras unstructured, linear, and 360 work. I've been a bit quiet as of late because I'm doing a test of exactly that for a client and my brain is fried between that and philmColor related work.
08-22-2020, 05:01 PM
One thing that is exciting to me about it is the standards that are already in place and are being proposed are actually quite good. Metadata directly added to the file about how something was graded and on what display is amazing. So if something was graded on a crazy 3000nit Dolby display, but viewed on an 800nit LG OLED, it can easily be mapped down by the display and still look very good. Even grading something on the XDR display then outputting in HEVC and playing back on an iPhone 11 Pro, everything translates down very well. I am curious what you think the brightness standard should be? 1000nits sustained and 1600 peak is quite nice, and realistically that is probably going to be the standard, but I think the dream for me would be more like 1000nits sustained 5000nits peak eventually has the standard for a reference display someday. Or just go all out and have 10000nits be the standard. I want HDR reference displays to right on the edge of damaging my retinas haha.
I love going back to older films that have been remastered in HDR now. Watching The Dark Knight with 1600nits peak is like a whole new experience, it's like a layer of dim film has been lifted off of it. I love how you get a sense of sharpness just from the contrast alone. Each time I have seen a display with proper color and uniformity and a double in peak brightness, it just gets better and better. Atomic Blonde is another one that really has a nice HDR grade. And Greyhound is one of the nicest looking bits of HDR footage I have seen. So much detail in the shadows and lots of tasty peak brightness. I believe that was shot on DXL if I'm not mistaken and man does it look nice.
What I am really desperate for is a proper 1000nit HDR on camera monitor or something close. That is quite challenging, but it is definitely a bit frustrating when I feel like my phone has a better display with better contrast ratio, uniformity, color gamut and peak brightness than my cinema camera monitor. I want an on camera monitor that feels like I am looking through an empty picture frame.
08-22-2020, 05:03 PM
Red Komodo 6k Auto Focus First Look
By
Brad Watanabe
08-23-2020, 06:08 AM
Not saying that the AF is a joke, I'm saying that without a proper face-tracking feature it will not be reliable for most types of AF related work. You can't have something looking for focus in the middle of a take or while doing gimbal work. So the AF system will be unused by most people until it has software that can hold that focus onto objects that are important in the scene.
The Canon or Sony AF system in video is nothing without their face-tracking and eye-tracking component.
The AF system should not be included as a key feature need when considering buying Komodo. Only what is present in the camera at the moment is what you should focus on, not promises of tomorrow because they might not happen. We might never get a software update that enables face-tracking, because we don't know if the processor can handle that or not.
So the important thing for Red is to at least "make room" for processing power to handle such a software, or else the AF system will just be unused in most cases. The potential of AF only occurs when we can shoot scenes wide open hand-held of an actor and the focus is rock solid on their face or eye. This is the whole reason why many don't get cine cameras and instead Canon or Sony DSLRs for their video work. A Red camera with similar AF function, even basic face-tracking in comparison, would really change the game for many shooters. So, in my opinion, either go all in and do AF with face-tracking or focus on other improvements of the camera instead.
What lenses would you use on such a 12K camera? It's unrealistic in my opinion. Most lens manufacturers have just released FF lens lineups and there are so few cine lenses for cinema work if you go larger on sensors from FF. Large format still photography is unrealistic for cine work and medium format, which is Hasselblad and 65mm require very expensive and special case lenses.
Better to do a new sensor, Vista Vision with 12K to get 8K distribution. But in my opinion, stick with 8K VV sensor and increase low light quality, DR and framerate cap. Going larger than FF is not a good choice and leads to tons of problems for video work.
A 65mm cine camera would be a great addition and competition towards the Arri 65, but I feel that's more realistic as a special case camera like the Panavision Millenium DXL2 for work related to IMAX releases. Because of difficulty having cine lenses for a 65mm camera, there won't be any realistic normal sales of these cameras, because it can't be scaled up. Full format video is easier because it's already a format that has a lot of lenses out there and the new FF releases of cine lenses are easier to do than going larger.
It's good to not stare blindly into "bigger and more" and think that's a good idea. There has to be a balance. Full format is the perfect balance between lenses and sensor sizes in terms of video. 8K is a perfect balance for 4K releases while 10 to 12K is the perfect balance for 8K releases subsampling.
So having a full format 12K sensor, 19-20 stops of DR, global shutter, 3200 noise free and eye-tracking AF that can feed focus to cine lenses, would be the perfect camera. Aiming higher than that instead of improving other aspects of the camera body and handling is just a waste of resources for something that's a diminishing return of improvements.
Personally I've not seen any actual benefit going higher than 6-8K. Noise handling etc. doesn't just depend on resolution and subsampling. Of course I see the better quality of the Monstro 8K over Komodo, but actual benefits, in my opinion, have to do with striking a balance rather than pure maximized power. Komodo in my opinion strikes a balance. It's not perfect, but close.
An 8K VV global shutter 100 fps capable sensor with face and eye-tracking AF and noise-free 3200 ISO inside a Komodo sized camera body with low latency video monitoring and wireless audio built-in, would, in my opinion, be the perfect cinema camera.
Aim for that, nothing more nothing less and we have a winner. Improving that further would be about keeping the same specs but getting it to be fan-less, low power with whole-day battery use etc.
08-23-2020, 06:44 AM
The Panavision DXL2 is a Monstro 8K sensor (40.96mm wide and 21.6 millimeters tall ), so not at all different in size than the DSMC2 Monstro and Ranger Monstro so I don't understand where your comment is coming from. Furthermore, nowadays, cameras don't need to have a 65mm sensor to be approved for Digital Imax Capture as the Sony Venice was approved a few months ago and has a sensor smaller than Monstro (36mmx24mm).
As for lenses, the following cover the Alexa65 sensor so a 54.12 mm x 25.58 mm image area :
And I'm pretty sure that Panavision has a bunch of lenses made for 65/70mm film cameras that would also work so lenses seems to not be an issue. (plus : on an RF mount you can adapt Hasselblad's primes with no problem at all)
EDIT : and at least every Christopher Nolan movie with shots shot on actual Imax cameras used Hasselblad lenses at some point so I think it's a testament to the fact that it's doable.
My point is : why IF that was an actual possible roadmap wouldn't we want RED to go in that direction. We already know that 8K is happening. Not everywhere yes, not tomorrow and certainly not in my home for quite some time BUT that doesn't mean that it's not coming faster and as you said, a 12K sensor would be great of 8K distribution (in fact, even 10K would be plenty good for that) and by making the sensor bigger, they can keep the pixel pitch closer to what they have on let's say the Monstro sensor. They are a Digital CINEMA Camera company and more and more high budget movies and TV shows are shot on yes full frame and VistaVision but Alexa 65 also. Even Parasite was shot on it. And is a visual masterpiece (and well worthy of every praise it got IMO).
I agree with what you're saying nonetheless about DR and noise. Those parameters are a priority to me BUT right now we're already at a level that exceeds what can be shown even on (current) Dolby Vision equippement so I'm not sure DR improvement is as critical as it was 10 years ago. It would be nice, yes, but doesn't seem as critical right now.
As for autofocus, YES it is awesome BUT I don't think it will ever be as primordial for cinema production because I strongly believe that we will always need the human touch and feel on focus pulling. Even the errors are artisticaly important. I mean the scene in Joker where he dances in the toilets was almost improvised and it works and a technical and artistic stand point because focus was pulled by a human trying to keep up with an actor and DP who were improvising the scene.
Last edited by Bastien Tribalat; 08-23-2020 at 06:58 AM.
08-23-2020, 08:16 AM
I haven't either, but only because I haven't been able to go higher than 8K.
I will say this: there is much to be said for framing a little bit wide and having the freedom to punch in up to 2x-3x when need be. There is also something to be said against framing too wide and believing that a 4x punch-in gives you the freedom to get multiple MCUs from a single master (wide) shot, especially when those MCUs are at opposite edges of the wide shot. I have really, really appreciated shooting 8K resolution masked down to 7K or sometimes 6K in the viewfinder and being able to recompose shots as I edit, just as I have cursed myself for doing too good a job filling an 8K frame and not really being able to reframe as a result. If I had 12K, I'd mask that down to 10K for composition and be very happy that there's a very solid 4K image anywhere within the 6K center of that frame.
08-23-2020, 08:51 AM
"Special case camera" means a camera that you only rent, you can't own. It's very hard to scale production of a camera that together with 65mm lenses becomes so expensive that there's no realistic point for anyone to own it without having a large rental business surrounding it. The focus here is to "own yourself". As long as someone isn't a billionaire that could easily spend whatever they want on camera tech, it becomes a pretty unrealistic camera to produce, except as a special case camera. Both Arri 65 and the Panavision DXL2 are rental only and that's where a 65mm Red camera would fit in.
Doable is not the same as practical or good business. Christopher Nolan produces the largest type of movies you can imagine, budget-wise. And that is my point, you fit the camera to where the market is and I don't believe there's enough of a market for these types of super expensive 65mm solutions, except in those cases like these large budget productions, which rent equipment because anything else is really just a waste of money.
Again, I think you misunderstand what I wrote. I'm not saying there isn't room for Red to compete with Alexa 65 and getting a 12K. I'm pointing out the maximum type of camera before it becomes irrelevant to max out even more. Where diminishing returns of image quality and sensor sizes runs out of reasons to grow bigger and "more". A 12K 65mm Red would be maximizing this and it would be a very special case camera. Very few would need it and renting it would be far more reasonable for when those projects arise. Remember, you would need a fully functioning post-infrastructure to also handle all that data. Your examples are productions where they rent those cameras, therefore, a Red 65mm camera makes more sense as a rent only, just like Arri 65 and the Panavision. You can't buy those cameras for yourself.
So when I speak of a "perfect camera", I speak of one that is ownable. Something that is reasonable to buy and own. A perfect camera for large budget production is one thing, a perfect camera to own without being a billionaire is another.
When the technology is there, when we have the high-performance cine lenses having internal focus motors and face/eye-tracking software locking focus in a scene, the industry will change. Of course, there are artistic reasons behind pulling focus manually. But there's been many artistic things that have disappeared over the course of technology evolving.
Most of the time, focus in a scene is pretty basic: lock onto that person's face. And the ability to shoot wide open and hit the focus in every take of a scene so that there's never that loss of good performance just because the focus puller didn't hit the mark, those problems are gone. And with internal motors, special software for autofocus in cine applications could incorporate a play area for a focus puller to remotely work in tandem with the autofocus; letting the autofocus hold a lock on minor pulls while the focus puller pulls the larger racks.
Because no one has made this into reality, there are no people to report how much it solved problems and challenges during a shoot. The potential that a cine-based AF system would bring to the filmmaking world is through the roof and seriously ignored.
08-23-2020, 09:22 AM
Over 100 years ago, the magazine "Puck" carried the quotation Things move along so rapidly nowadays that people saying: “It can’t be done,” are always being interrupted by somebody doing it.
Blackmagic has already thrown down a 12k camera. Tokina Vista Primes have the coverage needed (granted, circle of illumination is not the circle of definition).
Red Komodo has scaled cost down 4x from where it was 4-5 years ago...which means that a camera with 4x the pixels should be in reach of what people spend on DRAGON back in the day. Please don't tell us you don't want it because it cannot be done. Let's see what can be done, and then say whether we really want it or not!
08-23-2020, 09:44 AM
Of course, but the key here is the things that have physical limit or challenges. Vista Vision is in my opinion both maximized in what the tech cost and what the performance is. Focus pullers are already struggling with wide open FF lenses, so scaling from s35mm to Vista Vision is not the same as the jump from Vista Vision to 65mm/medium format. The challenges and problems that 65mm produces require a scale-up in other areas that is pretty impractical for anything but the largest productions. So even if things move forward fast, doesn't mean it's reasonable to use it, either through cost or performance.
65mm will be a niche system for a very long time, not because of cost but performance and complexity of shooting with it. Vista Vision as a sensor size format is really the max size that is reasonable to "own". And 12K, in my opinion, is the max resolution to reasonably shoot in. Remember that as long as CGI and real-time graphics/game engines drive resolution distribution together with streaming bandwidth, we won't see 8K for a long time as a wide adopted format. 8K TVs right now are useless because we don't have any 8K things to watch. 4K adoption didn't take off before game consoles started supporting 4K, but that also meant the games actually were in 4K. But the new PS5 and XBOX, while supporting 8K output, won't have any games that pull 8K resolution off. Previous console mid-updates like the PS4 Pro enabled 4K in every game that was released. That won't happen with PS5 or XBOX, developers aren't forced to support 8K resolution, so there won't be any high-end games going above 4K.
As long as 4K is the standard in console gaming, whatever TV manufacturers say, there won't be a jump to 8K broadcast or movies. Combine that with CGI work still working in sub-4K resolutions. The jump into 4K in the gaming world was a huge hit towards the progression of other technologies like RTX raytracing, it set back the performance a lot in order to support 4K. So the gaming industry won't jump to 8K anytime soon when they are trying to catch up to pure 4K with high end rendering in other areas.
Bottom line, 8K will not be a "thing" anytime soon because the driving force to adopt it does not happen at acquisition or TV manufacturing, it happens at the CGI and game engine level. That's where the consumer adoption happen and it will be at least 10 years before we'll see wide adoption in the gaming industry.
| « Previous Thread | Next Thread » |
