Otis Grapsas
Carbonite Member
That's what I was thinking when I wrote the response. Kubrick was not very productive and he also could afford to work this way. Roger Corman is probably a better example for young filmmakers
SCARLET / EPIC swiss army knife camera details announced
- Thread starter Jack Daniel Stanley
- Start date
That's what I was thinking when I wrote the response. Kubrick was not very productive and he also could afford to work this way. Roger Corman is probably a better example for young filmmakers
1080p is good enough for 200 million dollar Hollywood blockbusters so suggesting it's aiming low is borderline hubris. However if you think your projects deserve something better than the likes of varicams, F900's etc, I say more power to you.
Rubbersquare
Well-known member
again, i'll say our short which played on a big screen was shot 2k on an f900 and layed off to hdcam, then transfered to betaSP, looked AMAZING. people thought it was 35mm. *I* thought it was 35mm.
Mattykins
Cute Member
Well, basically everyone actually.
Film is transfered at 2k unless it is a visual effects shot, in which the transfer happens at 4k. So essentially - everyone thinks HD is good enough.
..really this isnt even taking into account that the film is then projected, and is never going to apper to the theatre patron just as sharp as the actual film frame itself due to dust and smear and focus, ect.. which is why a blue ray often looks much sharper and more detailed than the version you saw in the theatre being a direct digital transfer of the fillm frame. even though its dropping that smidge of resolution to 1080p. so yes HD is really fine for film purposes, and in fact i think that sometimes a blue ray is even too sharp. at a certain point you begin to lose that dreamlike quality that helps facilitate your consciousness to drift off into the world of the film. (like those 3D mirror samsung HD LCDs.. its like watching behind the scenes footage..and doesnt feel right for film at all,,sports..sure.. movie ..yuck) ..its the same for the 24p effect. if you view the same film shot at 24p vs 60i the 24p film will just feel more right and let you experience it better. so unless your going to imax or something as such, using a higher res than 1080HD or 2k isnt really necessary.
BUT.. and that a capital but.. if i had the choice i wouldnt see why i wouldnt shoot at a higher res and give myself options for reframing or cropping and zooming and some slight slop room in the post comp work, because on the downconversion its going to be briliantly sharp for output ..it even helps smooth out any noise/grain and gives you a little breathing room on the focus during production if your off a hair. so ..while shooting at higher than HD isnt necessary. it certainly has alot of advantages and can come in handy in many different ways, and i have a filmmaking rule about not making things harder on myself if I dont have to to acheive the same results. so if your shooting 2k or 1080 you always have to try and strive to keep things pristine at the res you are shooting vs shooting higher res you have breathing room in many areas. ..not to work more sloppy .. just to shoot more confidently. we all have enough to worry about on the set already. but knowing that if i later decide i wanted that shot a little tighter or framed up or down a bit ..i dont have to spend more time on set, or worry about if it will work out ok.
anyway . i definitely understand your point about HD being enough..because i 100 % agree. but i also agree that having more than enough is certainly a good idea, and if given the option i would have to refer to my filmmaking rule about taking the easier route to achieve the same.. or in this case possibly even better results by using a higher res for production and then downconverting to HD/2k for a great end result.
Prodigi Pictures
Carbonite Member
a
Last edited:
Otis Grapsas
Carbonite Member
Extra resolution and downsampling cannot provide an image with less noise because on a given sensor size you have to get the pixel area to be 1/4 the original area in order to have 2x horizontal and vertical resolution. You are limiting dynamic range in the process and increasing noise. The downsampling cannot make up for this disadvantage. 2x resolution is -2 stops in dynamic range because the well depth in photons is 1/4 and your pixels get 1/4 the light. In any imaging situation, scientific or creative, you have to design the system for the maximum pixel size that will cover the requirements. That's why many HD camcorders are low res or interlaced (interlaced summing is 2x effective pixel size).
ok, ..but what im talking about really has nothing to do with the imaging or resolution of the sensor. im talking about the apparent noise in the final image of the red. if a final given image ..say 2k is downsampled to 480p you effectively average out alot of noise resulting in a smoother image with dynamic range intact. however you average out nothing by staying at the same post resolution. so my point is that given the red which imho doesnt have any real noise issues @ 4K, anything potential noise there can be largely averaged out in post when downsampling to 2k or 1080p . vs shooting @ 2k or with a 1080p cam if you have any noise issues youd have to smear or loose res or blur to get rid of it which all effectively lower resolution from the true 1080p or 2k image desired.
so while i mostly understand what you are talking about with the sampling of the image through smaller pixels creating more noise according to math. i do understand that the 4k image of the red is incredibly clean, and assuming the larger sampling resolutions will be as well, youd have nothing to fear and everything to gain in terms of noise by shooting oversampled and downconverting in post in real world application of the red cameras. because 4k is already superbly clean. so downsampling that to 2k will only give you an even cleaner image , along with all the other benefits.
so while i mostly understand what you are talking about with the sampling of the image through smaller pixels creating more noise according to math. i do understand that the 4k image of the red is incredibly clean, and assuming the larger sampling resolutions will be as well, youd have nothing to fear and everything to gain in terms of noise by shooting oversampled and downconverting in post in real world application of the red cameras. because 4k is already superbly clean. so downsampling that to 2k will only give you an even cleaner image , along with all the other benefits.
Otis Grapsas
Carbonite Member
You can average the noise out to a degree but how is this an advantage compared to the half res system that doesn't even have that extra noise in the first place? The dynamic range would be intact but still 2 stop lower than a 2k system.
My point is that the RED image is clean, but a 2k Red would probably be as clean at 4x asa and have 2 stops of extra dynamic range. I believe some users might prefer that.
To put it in cinematography terms:
system A: 4k, asa300, 9 stops dynamic range
system B: 2k, asa1200, 11 stops dynamic range
Pick one. The engineers generally pick system B in professional products (like a sony $100,000 digital cinema camcorder) and system A in consumer products (like a $200 sony consumer digicam).
My point is that the RED image is clean, but a 2k Red would probably be as clean at 4x asa and have 2 stops of extra dynamic range. I believe some users might prefer that.
To put it in cinematography terms:
system A: 4k, asa300, 9 stops dynamic range
system B: 2k, asa1200, 11 stops dynamic range
Pick one. The engineers generally pick system B in professional products (like a sony $100,000 digital cinema camcorder) and system A in consumer products (like a $200 sony consumer digicam).
Last edited:
Otis Grapsas
Carbonite Member
To be more precise about this, halving the resolution can theoretically produce a sensor that is as sensitive. This is because 1/4 the pixel area gets 1/4 the light so it saturates to the same degree. A smaller bucket that fills with less water. But smaller pixels tend to have lower quantum efficiency and narrower spectrum so this never happens. The noise and dynamic range advantage also remains even with identical ensitivity because less photons can be gathered. One asa300 will be 66dB s/n and the other will be something like 54dB but the manufacturer can claim same sensitivity at 0dB even though it looks like 12dB of gain to the eye. Equal performance sensitivity will be two stops higher on the large pixel. Equal performance sensitivity = sensitivity when the output has the same signal per noise ratio on both sensors regardless of the gain setting required for that.
Dvader
Well-known member
That's what I was thinking when I wrote the response. Kubrick was not very productive and he also could afford to work this way. Roger Corman is probably a better example for young filmmakers
I actually think Stanley Kubrick was very productive
, not in the sense that he created dozens and dozens of films and allowed other people to do work for him, but he was constantly working on something, and in the end every film he made was top quality (excluding his first 3). In the latter half of his career his output dropped off but he had developed and worked on many projects in between the ones that were released. You know, he had an entire room that if you walked into it you would have thought, 'this is his private library', when in fact every single book was on Napoleon. So perhaps the question should be, productive at what? if the goal is finishing films? maybe not, but if the goal is working tirelessly in order to produce some of the greatest films ever? veryI never said it wasn't good enough, I said it is good enough, only I asked, 'who wants to settle for good enough?' If I'm writing a scene or shooting one I would never want to settle for good enough. There will always be times when you have to, even like the 200 million dollar blockbusters you mention, but the future looks promising and 28k will give the option to not settle for 1080p.
Like above, I never said 1080p wasn't good enough, obviously everyone thinks it is good enough, because it is. But again, who wants to settle for good enough if you don't have to? And if you don't feel you need the creative abilities of 28k, then shooting 1080p isn't settling, it's what you want.
And really I'm not criticizing the resolution of 1080p because we all know it's a fine resolution, it just that 28k allows for many new creative abilities. I don't quite need 28k projected, going out to 2k or 4k will be just fine, thats how the movie will ultimately be shown in theaters or on blu ray, but before I do that, working with the original 28k is a brand new creative tool.
Why was 2001: a space odyssey shot in 65mm when we all know that 35mm is good enough?
Honestly, this seems like overkill. You could use a 5D MKII and get exceptional results without having to shell out so much money to shoot "28K" for these purposes.
28K for VFX shots? I'm sure someone'll do it... but 4K has been incredible, 5K? Sure! It's really just not that necessary and I hope the people employing it are ones that will use it to benefit the overall end product. Say, Nolan and Pfister!
Scarlet S35 is realistic enough. Well, for me, at least.
Im interested in the scarlets because you have a cheap camera capable of shooting through larger sensors. Personally I say forget 28k, an extremely high resolution image does one thing, gives you an extremely high resolution image. But personally Im more interested in lensing cleaner with the larger sensors, that IMO is what RED has changed in video production.
Now if RED really wants to change digital video production forever they will create a new shutter that gives us something that looks and behaves exactly like shutter on a film camera, but with the niceties of a shutter on a video camera
Now if RED really wants to change digital video production forever they will create a new shutter that gives us something that looks and behaves exactly like shutter on a film camera, but with the niceties of a shutter on a video camera
Last edited:
Yes, the big brain on Red is 55 grand. Gotta be a better way to make a movie poster.
Anyone know what kind of computing power is necessary for 28k? Something from Cyberdyne?
Last edited:
Otis Grapsas
Carbonite Member
If you have enough memory and 100 times faster storage throughput, 28K will just take 100 times more to render vs 1080p and require 100 times more space.
The evolution of resolution will eventually stop. Depending on the application, an image covers a certain angle of the human viewing system that represents a limited resolution area. The resolution is low when the brightness of the image is very low (movie theaters). The resolution is also low in motion because humans decode only what is useful/interesting and the rest goes unnoticed. The high resolution area is a very small angle in the center of the field that contantly changes direction while decoding information which reduces true resolution in motion. It's our means to survive and protect ourselves.
I did a calculation a few eyars ago based on human vision facts and the total resolution for a static image and a normal human was about 1 gigapixel. At a wide aspect ratio and a 90 degree coverage, a 28K wide image would be just right I think the 28K image would be so sharp that it wouldn't have any dramatic value. I believe that the dramatic quality of projected cinema depends on the low resolution and low frame rate of the image because the lack of reallife sharpness and motion eases and relaxes the viewer into following the story. For most humans, especially those without training in film or engineering, watching a film is a dreamy state of removal from reality, it's like being 3 years old under warm covers and listening to a fairytale through the voice of someone you love and trust. I can't imagine that in 28k or 200fps.
The evolution of resolution will eventually stop. Depending on the application, an image covers a certain angle of the human viewing system that represents a limited resolution area. The resolution is low when the brightness of the image is very low (movie theaters). The resolution is also low in motion because humans decode only what is useful/interesting and the rest goes unnoticed. The high resolution area is a very small angle in the center of the field that contantly changes direction while decoding information which reduces true resolution in motion. It's our means to survive and protect ourselves.
I did a calculation a few eyars ago based on human vision facts and the total resolution for a static image and a normal human was about 1 gigapixel. At a wide aspect ratio and a 90 degree coverage, a 28K wide image would be just right
I think the 28K image would be so sharp that it wouldn't have any dramatic value. I believe that the dramatic quality of projected cinema depends on the low resolution and low frame rate of the image because the lack of reallife sharpness and motion eases and relaxes the viewer into following the story. For most humans, especially those without training in film or engineering, watching a film is a dreamy state of removal from reality, it's like being 3 years old under warm covers and listening to a fairytale through the voice of someone you love and trust. I can't imagine that in 28k or 200fps.
Fliques Osman
Well-known member
Very good point
And frankly a problem for those that keep studying the craft of film. Takes away some of the magic that got us interested in the first place.
Is it a good thing to see the puppet strings?
Zak Forsman
Major Contributor
my girlfriend is an editor and she is easily pulled out of a movie because her mind goes to the choices the editor is making. i imagine cinematographers find their mind wandering to thoughts of grain and composition choices as well. my mind kind of drifts between the popcorn i'm eating and the dude hacking up a lung next to me.
..k i do get what your saying, i do. my point is only that in working here and now terms, the red one @ 4k has already surpasses or rivals the high end 2k or 1080 cams for dynamic range and noise. and when downsampled ..its cleaner yet. so in the choice of existing cameras why wouldnt i choose all the other benefits of downsampling 4k as well as an even cleaner image than one that was cleaner than most other cameras to begin with. ..its just that the math may be one thing if you compare same sensor types with identical image processing and encoding pieplines, but the real world here and now images are another. due to the mysterium and redcode raw, ect. theres just more that goes into a final image than the math behind the sensor itself. ..anyway i very respectfully AGREE to your statements in general, but ill keep my proof in the pudding rather than the ingredients. ..now when youve got a working and compareabley priced s35 2k cam with useably greater dynamic range and less noise in its final output than we can presume from a new 16bit FF35 red scarlet/epic running @ 4k and downsampled, my tune will undoubtedly change.
Otis Grapsas
Carbonite Member
First of all, as I mentioned above 2k vs 4k refers to sensors of same size. If you 4k has the same pixel area with a 2k camera and identical technology, it will have comparable performance in CMOS and inferior if it based on CCD because CCD noise increases a lot when megapixels/sec go up. This is handled by CCD with multiple taps but lets not get into that.
Does the dynamic range of Red One compare to the better of the 2/3" camcorders or the more expensive digital cinema products? Beale's test shows a little less than 8 stops.
http://www.bealecorner.org/red/test-080108/pics/StoufferT4110-Red-20D-1.png
4000/20=200 which is less than 256=2^8 which would be 8 stops
I have heard about more than 11 stops but I think that would require a very short exposure in order to hide the dark current (which is e/pixel/sec) or a very cold camera (dark current is halved for every 6-9 degrees C you can drop).
Is there an actual test with a production camera at normal exposure that shows higher dynamic range than the 2k competition? A direct comparison at normal settings?
Does the dynamic range of Red One compare to the better of the 2/3" camcorders or the more expensive digital cinema products? Beale's test shows a little less than 8 stops.
http://www.bealecorner.org/red/test-080108/pics/StoufferT4110-Red-20D-1.png
4000/20=200 which is less than 256=2^8 which would be 8 stops
I have heard about more than 11 stops but I think that would require a very short exposure in order to hide the dark current (which is e/pixel/sec) or a very cold camera (dark current is halved for every 6-9 degrees C you can drop).
Is there an actual test with a production camera at normal exposure that shows higher dynamic range than the 2k competition? A direct comparison at normal settings?
Last edited: