Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

What are you exited about (DSMC)?

Roberto Lequeux

Roberto Lequeux

Well-known member
Joined
Jan 3, 2007
Messages
5,403
Reaction score
0
Points
0
Location
Los Angeles
For me, Mysterium-X and Mostro!

Specifically, I kept saying to myself that dynamic range was hands down the thing to be exited about the new sensors. I am starting to think now, however, that equally important will be the added color rendition... more colors!

I'd love to hear what someone more versed in the art of cinematography would say on the subject.
 
DR.

I've been waiting 18 years for a camera to replace my film camera. I think monstro will be it.

"Ohhhh but if I'm spending $7k on a good still camera that is competitive with film then I could also get a great HD camera. Ohhh but if I spend $10k on a good HD camera I could just rent a 35mm motion picture camera. ohhhh but if I rent a 35mm motion picture camera I could get..."

I could never find a camera that was worth the money until now. FF35 Monstro. As much DR as film. As much resolution as film. Shoots 30-72 fps. Is competitive with a high-end HD camera. Sold... just as soon as they'll sell one to me. And I think I'll be happy shooting it for another 5-10 years.

It's the same reason I couldn't buy a laptop for 5 years. "I want a tablet. But then it's really expensive. I want it to be fast. But then it's huge. I want it really small and light. But then it's useless." I finally found a tablet that was fastish. Smallish. and cheap. But it took a LONG time.
 
I think that some people are getting their hopes up too high regarding dynamic range. If it was an easy problem to solve, Canon or Nikon or Sony or Panasonic or any of the other big companies would have done it already. Canon and Nikon are locked in an epic battle of DSLRs, but even so, DR barely increases in tiny increments with each new camera. In ten years they have added maybe 1 stop, or two stops. So I am skeptical when people are talking about how a newer model will have huge increases in DR over the previous model.

The main thing I'm looking forward to is a compact (inconspicuous) camera that can overcrank and shoot RAW. Overcranking is incredibly important to me.
 
What I am most excited about is that I can have a camera for every application and any shooting style I need it for.
 
Tom said:
I think that some people are getting their hopes up too high regarding dynamic range.
Click to expand...

Agreed. What little I've read about sensor design convinces me that such things are hard. Really hard. Even a 30% improvement in dynamic range is enough to deserve some mad props.

Tom said:
If it was an easy problem to solve, Canon or Nikon or Sony or Panasonic or any of the other big companies would have done it already.
Click to expand...

I agree. However, there are also some of examples of the big boys doing pretty boneheaded things with dynamic range.

Canon, for example, literally deletes 4 stops of dynamic range from their raw files when set to ISO 25,600, for no reason. Try it. Shoot ISO 1600 (rated at 25,600) and compare with setting the camera to 25600. The noise will be the exact same, but one of them is missing 4 stops of highlight headroom for no reason. (The only time Canon got this right was with the 10D.)

Nikon and Sony have nipped variable pattern noise in the bud already, but all of Canon's camera still plagued with it. Canon hasn't even bothered using any of the well-known methods for abating it, such as building the sensor with large enough masked-off borders on all four sides. This pattern noise reduces the theoretically useful dynamic range (based on random read noise only) down to a much smaller practically useful dynamic range, because human brains are trained to notice patterns.

Tom said:
Canon and Nikon are locked in an epic battle of DSLRs, but even so, DR barely increases in tiny increments with each new camera. In ten years they have added maybe 1 stop, or two stops.
Click to expand...

I think it has improved more than that, but I agree with your emphasis.
 
An other thing that has me quite exited is being able to shoot slum-dog style, but with better everything. I loved CC of BButton far more than Slumdog, same for Batman... but I do love some of the things you could only do with a super, light super small camera that -in my humble opinion- closely mimic things the human body and head can do. Remember how exited so many of us were when we saw the first Scarlet? People couldn't stop about tight spots, crevices and corners... I am SO exited about that! Add a gyro or two and we are COOKING!

Daniel, it is always a pleasure reading your posts.

Back on Dynamic Range and Color rendition, will Mostro really match film? I suspect that there is a long thread on that somewhere, but what is the quick and dirty final word?
 
I've been waiting my whole life for camera and instant technology like this.
and these 3K, 4K, 5K memories .... just wow ... time capsule
some things aren't forever
 
Probably not all film. But if it is 13+ stops as they claim. That's pretty damn good. Definitely into the "Good enough" realm. 11 in my mind is right on the line. But think about this. To some degree Canon and Nikon are still in a retarded megapixel battle. RED is dissapointedly sort of chasing them what with the 6k monstro. But at the same time I think RED sees the value in larger pixel wells than Nikon or Canon. Print tends to care about resolution more than film. After all 2k is good enough for most features right now. So focusing on DR seems like a safer bet than someone who is shooting a 16x18 poster.

When we render for TV we render 1080p. When we render for print we render at 4k. (well technically we actually render at between 16k and 8k since rendering algorithms don't have low pass filters and supersampling is done at the pixel level. ;))
 
When you sample a pixel in 3D you have to take multiple samples of random points within the pixel and then average them together. Otherwise you get extremely aliased images.

The Optical Low Pass Filter on a CMOS sensor optically performs the same action. It slightly blurs the image so that each sensor well takes in an area of light instead of a single focused point.

So if you sample a pixel in a 3D renderer you're actually rendering that pixel (but slightly offset) multiple times. When we're rendering we tend to mostly use 2x4 adaptive sampling. So based on a number of conditions it renders out between 2x and 4x resolution. If you render out at 4k. It's actually rendering out a 16k image and downsampling so that aliasing is minimized.

http://upload.wikimedia.org/wikipedia/commons/5/5e/Antialiased-zoom.png

So when an agency asks for a 4k rendering to go to print we are often actually rendering the equivalent of a 8k-32k image which the software intelligently supersamples down to 4k. (If there is a super fine line for instance it might render that part of the image at 32k equivalent resolution but if it's just a low contrast area only sample twice per pixel (8k).)
 
None of it was sticking till you showed me this, then it all made sense!

Antialiased-zoom.png


Thank you very much for the explanation!

Well, actually... to be perfectly honest I have been avoiding asking this question for some time now. Where does 3D come in?! Does this have to do with the 3D LUTs?
 
I meant 3D Rendering.

We do VFX mostly for commercial work so my point of reference on what clients want resolution wise is based on that. We're also in the rather unusual position of delivering to Interactive, Broadcast and Print departments. Interactive doesn't usually want anything larger than 1k. Broadcast usually wants 2k. And Print usually wants 4k+.

If RED made a camera to produce assets for each requirement then they could create a super high dynamic range broadcast and interactive camera. But Nikon still needs to satisfy the print department's 4k+ and as such their hands are tied in creating chips with more DR (bigger sensor sites).
 
I feel like I need to create a diagram or something. :D

DR is inversely related to resolution.
Nikon, Canon are slaves to print resolution requirements (4k+) therefore they can't deliver the same DR as easily as RED since RED could theoretically deliver a 3k camera and very few people would complain.

(All of the above was in response to the comment that Nikon and Canon have only offered small incremental improvements in DR over the years. Aka Nikon and Canon are involved in a seperate specs war and bound to different market demands which is why I suspect they haven't been able to deliver vastly improved DR.)
 
Gavin Greenwalt said:
Print tends to care about resolution more than film. After all 2k is good enough for most features right now. So focusing on DR seems like a safer bet than someone who is shooting a 16x18 poster.
Click to expand...

If 2K is the highest resolution that needs to be delivered, then I think it makes sense to try to get 3K or 4K acquisition if possible. I agree that dynamic range, etc. are important, but I don't think dynamic and resolution are at odds with eachother.

Gavin Greenwalt said:
To some degree Canon and Nikon are still in a retarded megapixel battle. RED is dissapointedly sort of chasing them what with the 6k monstro.
Click to expand...

Unlike 99% of photographers, web sites, forums, and magazines, I don't think that the megapixel battle is retarded. I think the history of imaging sensors proves that a sensor with smaller pixels has the same dynamic range as a sensor with large pixels.

Gavin Greenwalt said:
DR is inversely related to resolution.
Click to expand...

As Graeme mentioned elsewhere recently, resolution can always be traded to get DR.

The model for "small pixels = lower dynamic range" is this:

"A single pixel, in isolation, when reduced in size, has less sensitivity, more noise, and lower full well capacity."

So far so good. In the case of a single pixel, it's true. Part two is where I disagree:

"Therefore, a given sensor full of small pixels has more noise and less dynamic range than the same sensor full of large pixels."

The briefest summary of my position is Noise scales with spatial frequency. A slightly longer model describing what I think happens with pixel size follows:

"The amount of light falling on a sensor does not change, no matter the size of the pixel. Large and small pixels alike record that light falling in certain positions. Both reproduce the same total amount of light when displayed."

My research and experiments bear that out: when small pixels and large pixels are compared in the same final output, smaller pixels have the same performance as large.

Spatial frequency is the level of detail of an image. For example, a 100% crop of a 6K image is at a very high spatial frequency (fine details), whereas a 100% crop of a 2K image is at a lower spatial frequency (larger details). Higher spatial frequencies have higher noise power than low spatial frequencies. But at the *same* spatial frequency, noise too is the same.

A high megapixel image can always be resampled to the same detail level of a low megapixel image. This fact is sometimes disputed, such as by Phil Askey in a recent blog post; however, it was thoroughly debunked:

http://forums.dpreview.com/forums/read.asp?forum=1018&message=30190836

http://forums.dpreview.com/forums/read.asp?forum=1000&message=30176643

http://forums.dpreview.com/forums/read.asp?forum=1031&message=31560647

There is ample proof that resampling works in practice as well as in theory. Given that fact, it's always possible to attain the same noise power from a high pixel density image as a large-pixel one. And it follows that it's always possible to get the same noise from a high resolution image as a low resolution image.

The "small pixels have worse noise" idea has become widespread because of the following unequal comparisons:

  • Unequal spatial frequencies
  • Unequal sensor sizes.
  • Unequal processing.
  • Unequal expectations.
  • Unequal technology.

Unequal spatial frequencies.

This is the most common type of mistake. To compare 100% crops from cameras of different resolutions is the most frequently-made error. This is magnifying one to a greater degree than another. It would be like using a 2X loupe to examine one and an 8X loupe to examine another. Or examining a small part of a 30x20 print vs. a wallet-size print. It's necessary to scale for size in order to measure or judge any aspect of image quality.

The standard measurements for sensor characteristics such as noise are all measured at the level of one pixel, and necessarily so. Sensitivity is measured in photoelectrons per lux second per pixel. Read noise is variously measured in RSM electrons/pixel, ADU/pixel, etc. Dynamic range is measured in stops or dB per pixel. The problem with per-pixel measurements is that different pixel sizes have different spatial frequencies.

Nothing wrong with per-pixel measurements, per se, but they cannot be used for comparison with sensors of unequal resolution because each "pixel" covers entirely different spatial frequencies.

Using 100% crops and per-pixel numbers is like comparing two lenses at different MTF frequencies. If they have the exact same MTF curve, but you measure one at 50 lp/PH and the other at 100 lp/PH, you will draw the incorrect conclusion that one is better than the other. Same if you measure one at MTF-75 and the other at MTF-25. (Most people do not make this mistake when comparing lenses, but 99% do it when comparing different pixel sizes.)

Pixel performance, like MTF, cannot be compared without accounting for differences in spatial frequency. For example, a common mistake is to take two cameras with the same sensor size but different resolutions and examine a 100% crop of raw data from each camera. A 100% crop of a small pixel camera covers a much smaller area and higher spatial frequency than a 100% crop from a large pixel camera. They are each being compared at their own Nyquist frequency, which is not the same frequency.

Unequal sensor sizes.

It's always necessary to consider the impact of sensor size. The most common form of this mistake goes like this:

  1. Digicams have more noise than DSLR.
  2. Digicams have smaller pixels than DSLR.
  3. Therefore smaller pixels cause more noise.

The logical error is that correlation is not causation. It can be corrected by substituting "sensor size" for "pixel size". It is not the small pixels that cause the noise, but small sensors.

A digicam-sized sensor with super-large pixels (0.24 MP) is never going to be superior to a FF35 sensor with super-tiny pixels (24 MP).

Unequal processing.

The most common mistakes here are to rely on in-camera processing (HDV, JPEG). Another is to trust that any given raw converter will treat two different cameras the same way, when in fact none of the commercial ones do. For example, most converters use different amounts of noise reduction for different cameras, even when noise reduction is set to "off".

Furthermore, even if a raw converter is used that can be proven to be totally equal (e.g. dcraw), the method it uses might be better suited to one type of sensor (e.g. strong OLPF, less aliases) more than another (e.g. weak OLPF, more aliases).

One way to workaround this type of inequality is to examine and measure the raw data itself before conversion, such as with IRIS, Rawnalyze, dcraw, etc.

It's important to be aware of inequalities that stem from processing.

Unequal expectations.

If one expects that a camera that has 50% higher resolution should be able to print 50% larger without any change in the visibility of noise, despite the same low light conditions, then that would be unequal expectations. On the other hand, if one only expects to it be at least print the same size and the same noise for the same low light, then that would be equal expectations. Such output size conditions are arbitrary and in any case does not support the "small pixels are noisier" position.

Unequal technology.

If you compare a 5-year-old camera to a 1-year-old camera, it will not be surprising to find the new one is better than the old one. This is often the argument used to explain why smaller pixels are always better. But in one sense, it will never be possible to compare any two cameras with completely equal technology, because even unit-to-unit manufacturing tolerances of the same unit will cause there to be inequalities. It's common to find one Canon 20D with less noise than another Canon 20D, even if absolutely everything else is the same. Units vary.

I don't think that means we should give up on testing altogether, just that we should be aware of this potential factor.

So that summarizes the reasons why I think the myth has become so popular. Here is some more information and example images about pixel density:

Noise scales with spatial frequency

20D (1.6x) vs 5D (FF) noise equivalency

S3 IS (6x) vs 5D (FF) noise equivalency

30D @ 85mm vs 5D @ 135mm vignetting / edge sharpness / noise equivalency

400D vs FZ50

40D vs 50D

A paper presented by G. Agranov at 2007 International Image Sensor Workshop demonstrated that pixels sizes between 5.6 and 1.7 microns all give the same low light performance.

http://www.imagesensors.org/Past Workshops/2007 Workshop/2007 Papers/079 Agranov et al.pdf

I'm sorry this post is so long, I did not have time to make it shorter.
 
Roberto Lequeux said:
Daniel you never said what you are exited about the most.
Click to expand...

The price! I've been making do with the XH-A1 and, since December, the 5D2, but I expect to replace them with the Scarlet 2/3" and RED mini primes. The price gives me access to many of the benefits of the RED ONE; chief of all: REDCODE. That's what I'm excited about most. And who wouldn't be? Most of the benefits of shooting uncompressed raw in one tenth the space? It's a no-brainer.

I will also tell you what I'm looking forward to the *least*: proprietary file formats. :( I'm not sure if I can get over the idea of being locked out of my own files. Being forced to use an SDK to get to my own moving images is very undesirable, but I might choose to just live with it.

Other RED ONE features I'm looking forward to are:

  • More control over DOF (compared to XH-A1).
  • High resolutions.
  • Interchangeable lenses.
  • Fewer aliasing artifacts.
  • Build quality.
  • No more prism block artifacts.
  • 120 FPS
  • Meaningful firmware updates.
  • Customer service
  • Inexpensive recording media
  • Exposure aids such as false color

Then, on top of that, all the DSMC features, including the dynamic range, modular design, stills mode, etc.
 
Great answer Daniel! I'm so looking forward to all the new DSMCs... For all the above reasons and more!

Graeme
 
I think the Epic FF35 will change the way I shoot both stills and video., can't wait!

As far as dynamic range goes. I hope it gets better but it's not a problem for me at the level it's at now. You just need to know it's limits.

If you think the RED's DR is too low, just go and try shooting some transparency film ;-), like we used to do for all our still shoots.
 
Why does a larger sensor with equally large sensor sites demonstrate more DR than a smaller sensor?

I would think the reduced heat, cross talk etc etc etc would all contribute to a considerablly cleaner signal on the smaller sensor with equal sized sensor wells.
 
You must log in or register to reply here.
Back
Top