DJ Joofa

RedLake is also there: (http://www.redlake.com/)

The old bond was fun but it was not at all true to the books.

With the possible exception of Timothy Dalton as is considered by many to be the closest of all on-screen Bonds to the character in the novels. His movie "The living daylights" is one of my favorites among Bond movies, and I severely dislike Pierce Brosnan's Bond movies.

And more -- try a few values in al and see what happens....

I did try a few and it seems like, as I said before, valid in hex (0-F) -> ASCII conversion ('0'-'9' and 'A'-'F' ranges). Is there more to it? It is a clever construct. I recall using things like that a long time ago when I was active in assembly language programming.

Joofa

This talk of assembly language and the old days led me to look at some of the stuff I did back then, and I found something interesting:

Here's a little assembly language puzzle for you. What does this code do?

 

cmp al,00Ah

sbb al,069h

das

Seems like those BCD/ASCII conversions. Right?

More amazing to the geeks on campus was my Amiga 2000. It had 8 meg of direct access RAM... OOOOOOOOH!

I loved the Amiga computer. There was this great image editing program called "Deluxe Paint", which seemed quite advanced compared to what was being offered on PCs/Dos at that time. I was so glad when finally Deluxe Paint was ported to PCs and used it extensively for my work.

I used Whitman's CHASM, and Borland's TASM, but probably an early version before the Turbo Debugger you describe. I mostly used the DOS debug program. And, of course, the IBM manuals with the BIOS source code, for the PC, XT, and AT. I even kept a shrink wrapped set of those.

I used Microsoft MASM and Borland TASM, and Debug a lot. Turbo Debugger was really revolutionary. Perhaps the first to show extended registers in 386 and 486 when these models appeared, which I think Debug would not show, and it would also let you set hardware breakpoints. IIRC, a powerful feature of the Turbo Debugger was that you could set a breakpoint in somebody's else code that I used for hacking a lot. Those were good days without any memory protection boundaries between programs so one could do that. And, when the control would return to somebody's code Turbo Debugger would sense that and breakpoint there and then I could do my hacking/debugging onwards. Very useful.

I have several original PC's and XT's in storage. My thinking in keeping them was that some day, they'd be like the Ford Model "T" -- the first of their kind to make it big. I used to write assembly language for them, modify the ROM BIOS, etc.

John,

You reignited some fond memories of working with PCs and assembly language. Those were good days. I really liked those books by Peter "man in the pink shirt" Norton, which eventually led to Norton Utilities. Its amazing how much more 640 KB of memory seemed to appear at that time. I really miss the Turbo Debugger by Borland -- the first of its kind I had seen that will even run a program backwards! That was like seeing those "time reversal" movies, where everything flies back to its place; seeing those memory variables getting back to their original state as the program moved backward was amazing. Did you ever try Turbo Debugger? And, then that original IBM PC manual that had the full source code of the ROM BIOS. Going through that code was very instructive. They stopped listing the source code for ROM BIOS afterwards.

The RED ONE (obsolescence obsolete... remember?) is the 1st camera in the world to offer a sensor upgrade.

 

Jim

Red is a great camera, but unfortunately, the above claim in not true. Among some other possible camera company candidates that I have to check, we have offered a modular design where we can upgrade a sensor in less than 10 minutes, for a long time, much before Red camera was even conceived in practical terms. Please have a look at the following image that displays one of our camera that offers this design:

http://www.djjoofa.com/data/images/camera.jpg

BTW, the above image is of a desk in my office, and you can see the size of this camera by comparing it with the mouse next by, yet still, it offers 30% higher resolution than the Red One camera.

So technologically, the RED may be regarded as impressive, but financially, what has actually been achieved here? :rolleyes:

Hi Keith, interesting numbers and experience you have, which gave me some historical perspective. I don't know how to do a proper comparison of Red with the cameras of the early 1990s. With the current league of cameras, Red is certainly financially attractive compared to quite more expensive digital film cameras. I have based my comments on Red's specs. On the other hand I also do realize that practically what I have seen of Red footage in a few films such as "Knowing" and "The Book of Eli", was not as impressive as their specs say. I think Red has some reservations on how to do a proper "digital projection", perhaps at "real" 4K. I am hoping that if I am able to go to Red Day in Vegas this April, about which I'm not fully sure right now, I might be able to see some "real" projection, if they present it.

Having been involved in making advanced digital cameras everyday for the last several years if I have to use one word with Red camera I would say it is a great camera. I don't think Red's marketing was excessively hyperbole. I have seen more confusion among the cinematographers regarding the use of digital terminology associated with Red, but in the technical quarters what Red has advertised has always made sense.

However, I do think that people at Red don't fully understand the effect of incorrect language in their marketing, which they use often, on the widespread usage of certain terms. Terminology such as "linear light", and "new color science" are all technically incorrect usages, but it is becoming widespread among cinematography community due to lack of rigor by Red.

I can't help thinking that something similar will happen to the RED: They will open up a market and prove its validity, only to have it taken away from them, by more experienced operators, rather like IBM and their PC....

It would appear that Red is more of an "integrations company" and not a massive R&D company, so it should be possible for others, perhaps smaller companies and not necessarily big companies such as IBM, etc., to take on Red and bring competing products. It is certainly doable.

Even though there are some really beautiful digital images out there, digital hasn't found terra firme. Every day numerous posts surface, in various forums, that consistently highlight new issues which 'has-just-been-found' with latest digital cameras.

If by "digital" you have Red One in mind, then, Red is not a good model to take a sampling of the advancement of digital imaging, which is a happening in fields other than digital cinematography. Red could have been a disruptive technology, however, it chose to narrowly define its boundaries, working with a philosophy of a "better mouse trap" than film can provide. Monetarily Red is a good bargain. However, technologically, IMHO, it is more of an "integration company" than a massive R&D company to throw its weight behind the state of the art advancement in digital imaging.

This is one of those strange subtle places where things start going in the wrong direction. Silicon chips don't *have* ISO sensitivities. ISO is defined in terms of film curves and densities and such. There's no definition of the term for any other imaging system.

ISO definition is confusing for a digital camera and sometime a distinction is based upon ISO of a camera set and ISO relating to an image sensor. For camera ISO definition is recommended to be the one given by (4) below and for sensor it is advised to use the one given by (2) below.

The basic definition of ISO in digital imaging is:

(1) ISO = 10 / H,

where H some suitable chosen "adequate" exposure. For ISO 100 that would mean H = 0.1 lux-sec.

However, it is not always clear how to properly interpret the "adequate" exposure H and there are different ways to interpret it.

(2) Definition of ISO based upon saturation is based upon that exposure when image highlights are just below max possible saturation signal. H now is defined as 1/7.8 of the exposure at the saturation point. (7.8 = ratio of 141% reflectance to 18% gray.)

Therefore now,

ISO = 78/ Hs

(3) Definition of ISO based upon noise is when the exposure generates a "reasonably clear" picture at around 40 SNR.

(4) Definition of ISO when the exposure generates a picture of medium output level corresponding to 0.46 times the max output. 0.46 = (18%)^1/(2.2 sRGB gamma).

I personally think calling Red One a digital film or video camera is fine as we are already inundated with technical jargon. For e.g., is it pixels or sensels on a Bayer CFA? The context makes it clear in most circumstances that Red is not a film camera and how it differentiates with "regular" video cameras that output to standard formats. Introducing new terms such as "data cameras" are perhaps only going to add to the list of technical terminology. If tomorrow we have a camera that outputs more color channels than 3 or 4 then shall we have more terminology to define that camera: tri-color camera, penta-color camera, octa-color camera?

I was amused to log in today & see the following message

 

"You have been banned for the following reason:

No reason was specified."

I was going to say "Welcome to the club", which includes distinguished alumni like myself that got banned from RedUser, but I noticed in further reading of this thread that you were restored. I guess you are high profile. ;)

Any list that does not mention "Crouching Tiger Hidden Dragon" is invalid :lol:. Further, Tom, imho, Terrence Malick has been a great letdown in "The New World".

Nope, it's Nyquist. To resolve N lines, you need 2N samples or more.

Unfortunately the way sampling theorem is written almost always, it does not make it obvious that the above is a strict requirement when dealing with a low pass signal. If you are sampling band-pass signals then the minimum sampling rate may be less than 2 x max. frequency present, if a certain relationship is followed -- otherwise schemes such as wavelets would not have their appeal as much as they do. (All bands in typical wavelet configuration are bandpass except one and hence it is routine to downsample them consistent with the fact that a bandpass signal may be sampled at a lower rate than twice the max. frequency.)

Canon, Kodak, Sony, Fraunhofer etc. might disagree.

 

However everyone is entitled to an opinion which doesnt need to be based on facts.

Von Krogh, you need to read carefully. I wrote "image quality" and not necessarily semi-conductor design of only the sensor chip. I am talking about full end-to-end image quality, which if you have been paying attention to my posts would have found out that I have repeatedly said that image quality is more than just sensor, where, unfortunately most discussion seems to focus solely upon, and rests on other stuff including but not limited to such things as electronic circuitry external to CCD/CMOS.

Have you ever personally investigated the results of even such stuff as where to put the ADC on an electronic circuit board to reduce the image noise (the sensor is not even the picture here in this case)?

I think all cinema cameras are moving to the CMOS sensor.

The gap between CCD and CMOS cameras is closing. However, using certain magic circuity external to the sensor, the CCD image should result in a higher image quality than CMOS, in many situations. The race towards CMOS is for a variety of reasons, and some of the early hype of 90's has been found to be incorrect. Moore's law does not hold really in its true sense in the image quality world. Economies of scale is not relevant in the typical sense here. Please see Wong's IEEE paper.

The key to a superb image quality is as much in the circuity surrounding the sensor, as the sensor itself, however, CMOS may not offer that luxury all of the times that a CCD does in many cases.

Probably the best low light camera in the world right now is the Sony F23. Superb latitude, and an astonishingly low amount of noise. Certainly not the cheapest, but the best, by a considerable margin.

Hi Mike,

"n the world"??????

Perhaps F23 may be the king in the filmmaking world as you seem to suggest. I have never used it so I don't know. However, we have developed some amazingly low light cameras, with some even greater than 4K by 2K resolution, that are not in the domain of filmmaking, using some very advanced processes. Unfortunately, the big players in film-type cameras have done some very routine run-of-the-mill solutions that limit making really low light cameras, with ultra low noise. I was sincerely hoping that Red would do something along the lines as it like to touts itself as "the camera", but unfortunately, not so .....

Three phase motors are much more cost effective than single phase -- no capacitors or starting crap.

Hi John, a small refresher :rolleyes:, not all single phase motors have a capacitor for starting, some that come to mind right now are split phase motors that produce torque by having 2 stator windings; shaded pole motors; reluctance motors, and hysteresis motors.

Algorithms my boy! It's all done with Algorithms. If you have the right algorithms you can do anything:-)

Nope, David is right. A 200% blowup would suffer in sharpness and would appear blurry.

just that the 'nirvana' folks expect with HD is more a tune to 525 TV just with twice as many lines. ... in fact they don't really have much more than they did when 525 TV gave you 280 lines.

Unfortunately, a fundamental premise of HDTV has been ignored in comments such as above, and in many audience surveys. The typically larger screen size of an HDTV allows an enhanced image portrayal over standard definition different than mere physical resolution. A standard definition image can result in a pleasing image as long as the camera operator fills up 525/625 line system appropriately. However, an HD camera operator has a different choice while keeping more or less the same "observed resolution" as an SD system by altering the field of view (framing). In this situation, from the same viewing distance away, both SD and HDTV offer similar observed resolution, however, a totally different viewing experience is provided for the user: the HDTV shows more of the scene, more of the total scene action and objects. Thus, the HDTV offers a radically different perspective than SD, and a way of presenting image content that capitalizes on freedom in framing wider and more true-to-life angles of view, and providing a greater sense of reality.

Come on, DJ. You can't just leave that hanging there. ;-) What movie? Who did you play?

Hi John,

I am not too pleased with my performance so we shall leave it unidentified :rolleyes:

... a lot of HD photography can rival some 35mm photography for sharpness. But I was expecting something a little more detailed from a 4K RAW sensor camera. I also have to admit that after recently looking at some Super-35 footage I shot go through a D.I. that I was also somewhat disappointed with that level of sharpness too, so maybe my expectations are just too high -- I guess I want everything to look as good as 35mm anamorphic photography does.

David, digital frequency and analog frequency are two different things, and are related by the sampling interval. The notion of detail when data is in digitized form (just a time series) must be interpreted carefully, as HD/2K/4K have little meaning unless sampling density and scene framing are considered. When you have data in computer it is just an array of numbers and the notion of sampling interval is not directly apparent. For e.g., a 1920x1080 grid with alternate dark and white pixels give the same max digital frequency of 0.5 and a 4K sized grid with alternate dark and white also gives the digital frequency of 0.5. Which one has more detail?

I think Red has a pixel size of a little over 5 microns, so it sampling density is more or less the same as about any reasonable HD camera out there. On the surface it appears like a conundrum? How can Red claim being a "higher detail" camera than HD when it sampling density is more or less the same. The answer lies in the framing of the shot. Supposed an HD camera and a Red camera are identically framing a shot, as it would perhaps be a case with any cinematographer, as they would like to frame a shot considering scene objects, and independent of the resolution of the camera. Under identical scene framing, which would mean that one would have to use a longer lens on a larger sensor sized camera (Red), the scene is "magnified" in terms of sampling grid, since the sampling density has not changed. The result is tantamount to saying that as if Red's sampling density has increased, though physically it has not. Therefore, the same max digital frequency represents a higher analog frequency now, and hence, sampling of higher detail.

Therefore, the perception of detail is also relative, and unless sampling density and scene framing are known, no direct comparison of higher detail in terms of analog frequency can be made from just looking at digital data.

Unless you can persuade Tim Tyler to set up a: "Cameras that Fanboys are Currently Starching their Shorts Over but Which Don't Actually Exist Yet" folder, kindly take your evangelising about what is effectively vapourware to somewhere where it will be appreciated, if such a forum exists.

Unfortunately the didactic contribution of online forums is not being properly appreciated in the comment above. There is always an opportunity to learn even from fan boy sites if well-meaning and knowledgeable contributors are present. Before, I discovered RedUser and Cinematography.com websites last year I thought I knew most things about digital cameras; after all I have taken courses in filmmaking for a long time, make short movies, have been for several years and still develop high end specialized cameras by my own bare hands everyday, and even acted in a major Hollywood movie :lol:. However, I realized there were holes in my understanding of some issues in digital camera and cinema and took that as a learning opportunity to overcome it and am very grateful to such sites in furthering my knowledge.

Is Red One, and more importantly the new products from Red vapour ware? Though, I personally think that Red has put itself in a difficult position of overpromising, and so any hint of under-delivering can hurt its reputation, that has not hampered me from sifting through the information available from Red online resources to develop a better understanding of how an advanced digital camera should be developed.

Can Red be improved so that it is no-longer a "vapour ware"? Certainly, as I personally don't think that the best approaches were taken in its development, and some "run of the mill" type stuff in camera production has also made its way into this camera that is being touted as "industry changing." However, that shall not impede in appreciating what Red's intentions are and what can be accomplished.

I'm not sure if that color to the noise is partly if the debayering algorithm is assigning a color to that noise. My guess is that noise comes in all colors.

"Color to noise" is being assigned by the optical filter array and not the debayering algorithm; though, as John Sprung has eloquently pointed out that the term "noise color" is being used incorrectly here. White noise is uncorrelated noise as John has explained and not to be confused with the "color" imparted to the noise by the filter array.

The implication of this is that you might as well store 8 bit and diffusion dither rounding errors with Gaussian noise during postproduction

Not the right approach. Adding noise after quantization is not as effective as adding noise before quantization.