Jump to content

The cheap RED aesthetic.


Recommended Posts

  • Premium Member

There are still people who prefer the sound of crackly old mono '78 vinyl to pristine modern stereo recordings

 

78 RPM records were almost all phenolic rather than vinyl. Vinyl started with 33 1/3 RPM and 45 RPM in the 1950's. They were called "unbreakable" because vinyl is flexible, while phenolic records break about like a ceramic dinner plate.

 

Analog mastering early in the vinyl era was really quite good. Compare the best Caruso you can find from 1907 with Dave Brubeck's "Take Five" from 1957, and anything you want to pick from 2007. It should be very clear which half century produced the greatest improvement.

 

 

 

-- J.S.

Link to comment
Share on other sites

  • Replies 103
  • Created
  • Last Reply

Top Posters In This Topic

  • Premium Member

I have said it once and I will say it again. It is impossible to capture natural looking images with digital sensors that incorporate pixels that are arranged in a tile mosaic pattern of square boxes. Film with its random grain pattern more closely approximates they way that the human eye sees reality.

 

The eye is a filter. We don't want to reproduce that filter in our camera systems, because the result would be like hanging two 85's when you really need one. Ideally, we'd want to reproduce the range that the eye can see, and let it do the rolling off. We see *with* our eyes, not by looking *at* our eyes.

 

The advantage to a random sampling structure like film grain isn't that it's in any way like the human eye. The advantage is that if you let it alias -- and it does alias -- the result is random and subjectively less objectionable than the moire patterns that come from cheating Nyquist on a regular grid of photosites. The success of digital still photography proves that with enough photosites and a correctly designed OLPF, the little box thingy can work.

 

 

 

 

-- J.S.

Link to comment
Share on other sites

  • Premium Member

78 RPM records were almost all phenolic rather than vinyl. Vinyl started with 33 1/3 RPM and 45 RPM in the 1950's. They were called "unbreakable" because vinyl is flexible, while phenolic records break about like a ceramic dinner plate.

It's true some 78's could be smashed like that, but most had a sheet of paper moulded into them that prevented it. Many cracked 78s can thus still be played, and the resultant clicks can be removed with audio restoration software. I've done a few of those myself, and the results are quite worthwhile.

 

Actually most the early discs were made from shellac mixed with powdered gypsum, wood flour and other "secret" ingredients purported to lift that particular manufacturer's product well above the standard of their low-life competitors! Phenolic (Bakelite) records were supposed to give superior sound quality, but they didn't last as long. The problem was that most 78s tended to be played on wind-up acoustic gramophones which were very hard on the records, since all of the sound had to come from the needle vibrating in the groove. Some manufacturers experimented with "audiophile" 78s that were only intended to be played by lightweight electric pickups, but they were superseded when vinyl LPs were introduced in 1948.

 

We also need to get our facts straight here. While there a few eccentrics (or ratbags, depending on the state of their finances :lol:) who insist that 78RPM shellac disc are the only TRUE path to acoustic verisimilitude, the bulk of the market for 78-originated music is people who appreciate that recordings made on 78RPM wax discs are the currently the ONLY means of hearing performances of musicians who are either now dead, or past their prime. They generally don't need to have their actual copies on 78s; vinyl, tape, CD or even MP3 copies are all OK for most enthusiasts.

 

A possible future technology would be a system that can analyze old recordings and essentially reproduce the entire performance using synthesized instruments, in the fashion of a player piano. Some major software advances would still be necessary, particularly in the field of producing truly convincing synthesized instruments, but there's nothing to say it can't be done. There would certainly be money in that for the record companies.

Analog mastering early in the vinyl era was really quite good. Compare the best Caruso you can find from 1907 with Dave Brubeck's "Take Five" from 1957, and anything you want to pick from 2007. It should be very clear which half century produced the greatest improvement.

-- J.S.

Actually, even though the CD was introduced in 1981, it was only fairly recently that 24- or 48-track analog mastering began to lose significant ground to all-digital audio post-production.

Originally in the 1980s Sony thought they were going to clean up re-equipping recording studios around the world with their new range of all-digital equipment, but it was ridiculously expensive, and most consumers seemed perfectly happy with analog-mastered CDs, so there were very few takers. (The CD format certainly has a number of advantages over vinyl or tape, but there's no law of Physics that says users have to take advantage of all of them!)

 

The real reason all-digital production has come to predominate is simply that it can now be done much more cheaply on inexpensive mass-market PCs.

Link to comment
Share on other sites

  • Premium Member

I find something as simple as dialing up Saturation to 1.2 or 1.3 in camera (yeah, I know it's meta-data) has a much, much more pleasing image and skin-tones than if you leave it at 1. I figured out why the first few projects had grey skin-tones once I did...

Link to comment
Share on other sites

  • 7 months later...
  • Premium Member

Hi Vincent,

Interesting post. I look at both film and RED originated content that comes through Cinelicious all day long. I have to agree I can usually spot RED footage right way.

 

David... I just saw Winters Bone and loved the story & acting but found it over graded and the horrible clipping very distracting. I'm not sure if you can chalk clipping up to bad production value. We've also recently been able to work on some MX footage on a few commercials. Maybe a bit more latitude but still feels sort of plastic-y. I'll look forward to checking out what you do with the MX sensor.

 

I'd also like to say that RED isn't much cheaper than film. Vince... have you're producer call us if it's a price issue why they can't shoot S16mm. We just did the DI on an indy feature where the DP and director wanted to shoot S16mm and the producers wanted RED. We were able to come up with a scan-once workflow (finish looking "dailies" from Spirit to ProRes 4444 - edit dailies, conform-free, file-based finish touch-up w/output to SR for festivals). This workflow brought the cost down from a traditional film DI and they were able to convince the producers to shoot film... Anamorphic S16mm which is rare and looks really cool (SR3, Panavision lenses).

 

Best,

 

Paul

 

Paul,

 

I agree with your assement of "Winter's Bone" completely. Just finished watching it with a friend. I pointed out the clipping to her and some of the grading, esp. the day for dusk. She saw them right away and didn't like them either.

 

I definitely felt that a more graceful roll-off would have helped a film like this to bring out the odd beauty of the production design and locations.

 

BTW, a documentary S16 film that does exactly that in similar locations is "Serching for the Wrong Eyed Jesus."

Link to comment
Share on other sites

You mean this?

 

red1l.jpg

red2g.jpg

red3p.jpg

 

Instead of this?

 

001vnd.jpg

 

It's the camera but also the processing and most importantly agressive color correction applied on poor input.

 

RAW cameras take much of the control from the cinematographer and give it to the color corrector. The DP that uses a RAW camera cannot really rely on a look provided by film stock or even a standard set of matrix/gamma provided by the camera manufacturer. While good old timing provided limited control, digital RAW provides way too much power. Add the camera source problems that need to be corrected and you have a real problem.

 

I'm surprised we don't get many fist fights over color correction with the DP control of the creative intent lost to such a degree. I guess a DP could learn the digital image development and color correction tools and handle the color correction directly, but what are the chances of that happening? The tools typically work in the wrong domain and distort the color space further because they attempt to fix images using the wrong approach (3-way to fix color space issues and so on).

Link to comment
Share on other sites

Otis, I couldn't agree more. For some reason people think, because they have a supremely color-correctable image, they should correct the hell out of it. I have seen some nice, natural looking footage from Reds just massacred in post. The option is there to grade the image to a more natural look, but 90% of the time it's the opposite that occurs – going with jarring colors and an artificially sharp image. This current trend makes me sick. I wish the film mentality of getting it as close as possible in-camera and making only small tweaks later was used more often.

Link to comment
Share on other sites

  • Premium Member

One of my first posts here was about this very topic. Can't find it, I guess the archive doesn't go too far back? "The Waitress" and "American Splendor" motivated the post. There were colors and intensities all over those films that are just not possible to create with real lighting. Mattes/secondary CCing can make this really jarring. Many times it works, and even with "American Splendor" (which has comic books woven into it's storyline) it can be justified. But I fully believe it can become rather tiring for the viewer. It's like MSG or high fructose corn syrup or walking along Universal City Walk at night.

 

 

P.S. I found the thread I mentioned above: http://www.cinematography.com/index.php?showtopic=31034

Link to comment
Share on other sites

Technically, digital is already significantly better than film. Although I've only used the original red, and I had some issues with it, I have no doubt the new one trumps super35 in terms of resolution, low light sensitivity, perceived sharpness, etc. And the promised HDR mode with the new model will surpass anything for exposure latitude. With still images it's the conventional wisdom that 135 film is similar in image quality to 3MP digital (maybe a little better) and full frame digital is "good enough" to replace 4x5 in most instances, and certainly to replace 6x9. Digital cinema will soon surpass IMAX on technical terms. Technically, digital has arrived; it's the filmmakers' turn to adapt, as still photographers already have. Once the status quo shifts to digital cinema it will shift in a major way and the obsession with "film look" etc. will just disappear. We won't look back, for better or worse. No one would shoot Black Narcissus, Best Years of Our Lives, or Barry Lyndon today; no one will shoot like Shutter Island tomorrow.

 

That said, I think the original Red has poor skin tones and a very distinctive look to it, a kind of plastic magenta/cyan cast, that's certainly unique to the camera. I'm not a huge fan of the look but it's not terrible (especially when mitigated in post) and it seems to be less present in more recent footage. Just by virtue of this look being unique and associated with the lower-budget productions shot on red, it's come to look "cheap." The company is a bit of a victim to its own success. Pirates looks amazing because it's shot great with good production design and also because it has incredibly in-depth color grading. It's hard to know how close you could get to that look without a huge post budget, but even the cheaper current red productions look a lot better to me than they used to.

 

Also, this is heresy since both cameras have poor image quality, but the hvx and 7d both have great color. Ungraded 7d footage, when not plagued by technical issues, looks quite nice and the skin tones are remarkably good.

Link to comment
Share on other sites

  • Premium Member

.... I don't think one can say that digital has surpassed film if it can't handle skin-tones and over exposure well, which I've never seen a digital camera do, and highly doubt one ever really will do. Resolution is perhaps the most unimportant "number" around. I can have a 15K or whatever image of black and white or abhorrent greens ect and yes, it will resolve well but it'll still look horribly digital. There is a reason many people shoot with diffusion on digital cameras (as well as film).

Link to comment
Share on other sites

Does this cheapness apply to Mysterium, or Mysterium X, or both?

 

I will agree that no one has really used the Red One well in a feature, as of yet. Not that I have seen, anyway. But it's just a matter of time. The camera is perfectly capable, especially the MX.

 

Tom, welcome back!

 

R,

Link to comment
Share on other sites

 

It's the camera but also the processing and most importantly agressive color correction applied on poor input.

 

RAW cameras take much of the control from the cinematographer and give it to the color corrector. The DP that uses a RAW camera cannot really rely on a look provided by film stock or even a standard set of matrix/gamma provided by the camera manufacturer. While good old timing provided limited control, digital RAW provides way too much power. Add the camera source problems that need to be corrected and you have a real problem.

 

I'm surprised we don't get many fist fights over color correction with the DP control of the creative intent lost to such a degree. I guess a DP could learn the digital image development and color correction tools and handle the color correction directly, but what are the chances of that happening? The tools typically work in the wrong domain and distort the color space further because they attempt to fix images using the wrong approach (3-way to fix color space issues and so on).

 

Last summer I shot a feature on the Red MX.

 

I just completed color grading the film from the original r3d files.

 

My thoughts: After grading about 2500 shots...

 

The camera produces incredibly detailed and noise free images.

 

However...

I feel that the camera has an "odd" color look to it. I've played with these files every which way, and while they can look stunning, it's difficult to make the images look like they did live on the set. Maybe it's the RAW converter, maybe it's the imaging chip, I have no idea. I do know that still images shot RAW with my Canon 5d look normal when processed in Adobe Camera RAW. The RED R3d files need some unusual tweaking to look "photographic" for lack of a better word.

 

Right after I graded the RED movie, I graded a project shot on the Sony f-35. Right away the images looked the way I imagined, though they have a bit more noise and are not as detailed as the RED images. But, color wise, they are much more "natural" looking. The Sony seemed to have about the same dynamic range as the RED MX, but does require a bit more light for best exposure.

 

As for the DP loosing control of the process? Well, I've taken back control on these projects:) I had a choice between sitting and advising a colorist or spending a little more time and doing it myself. I chose the latter for these projects.

 

I'll also add this: One sure learns the characteristics of the camera when one grades 2500 clips oneself!

Link to comment
Share on other sites

Technically, digital is already significantly better than film. Although I've only used the original red, and I had some issues with it, I have no doubt the new one trumps super35 in terms of resolution, low light sensitivity, perceived sharpness, etc. And the promised HDR mode with the new model will surpass anything for exposure latitude. With still images it's the conventional wisdom that 135 film is similar in image quality to 3MP digital (maybe a little better) and full frame digital is "good enough" to replace 4x5 in most instances, and certainly to replace 6x9. Digital cinema will soon surpass IMAX on technical terms. Technically, digital has arrived; it's the filmmakers' turn to adapt, as still photographers already have.

 

Cine film has slightly other properties than still film and works in a different environment with different requirements, grain doesn't degrade moving images as much, films are of higher quality (new-gen negative-film without orange mask) while digital has to deal with some issues that aren't apparent in still photography like cooling, movement artifacts, data storage, lack of ability to deal with artifacts in post (like aliasing).

 

3MP digital vs. 8perf 35mm is typical for consumer-scanned negatives, the equivalent to all the lousy 2k-telecines from features shot on not too well-exposed Kodak 5219 (even for daylight) - this kind of 35mm-cinematography makes the digital transition easier, that's for sure... But it's not what 35mm is about and capable of.

 

The ARRISCAN is an interesting example "between the worlds" and gives an idea what digital has still to deal with, because it incorporates the same sensor as a camera (D-20/21) but still has distinctive differences which allow it to work with higher quality:

-It doesn't use color-filters in front of the sensor but custom-made LEDs -> no color issues like in digital acquisition

-it doesn't use an OLPF

-no bayer-interpolation (three times the color resolution)

-micro-scanning, taking 4 images from each frame at different positions to generate true 6k-files which can be downsampled to aliasing-free, high-MTF 4k-files

-double-exposure to generate true 16bit files with high dynamic range and little noise

 

So this scanner has to merge 24 (!) exposures to get a high-quality file from film. Still, the transition chemical->electronics generates a huge loss that isn't necessary in digital acquisition - but digital has still to deal with some basic issues affecting IQ.

 

Being purely "objectively" is difficult, but I give it a try for very basic properties:

 

Resolution: modern film has 30%-50% contrast at 80lp/mm, that are 4000 lines (RED = 0%) horizontally in a S35-frame, it goes way beyond that and that's important for scanning (aliasing, MTF) but not very usable regarding contrast and doesn't end up in the master positive or in the DI-files (6k->4k). 2000 lines with good contrast and about 3000 lines extinction resolution with low contrast are realistic in a real-world DI - we are talking real resolution here, in the final files, taken with real lenses, going through the whole process - RED uses color interpolation and a strong OLPF with little to none oversampling to compensate (4.5k->4k, 5k->4k), on test charts (high contrast, black-white - ideal for digital) the RED 4k never was able to transfer usable contrast beyond 2500 lines. I'm well aware that RED claims otherwise.

 

Dynamic Range: the sensitometric curves for Vision3 shows a DR of 12-13 stops, counting the "shoulders" (film doesn't behave linearly at the ends) you can end up with 16 stops like Kodak claims - so it's not comparable to linear but clipping digital by those simple numbers. An HDSLR has about 9 stops, a RAW-file from a DSLR about 11-12 stops, the RED MX 11 stops and the Alexa measured under the very same circumstances up to 14 stops. They've done a 1:1 comparison (Kodak 7219 with ARRISCAN vs. ALEXA) for evaluation in Germany (switching public TV-production from S16 to Alexa) and they claimed ALEXA performed as well as the DI regarding dynamic range with less noise - which is sensational for digital acquisition, that's what the industry claimed for over a decade! But the ALEXA is a HD-camera (although with little artifacts and high MTF), pushing it to 6k (for 4k output) would compromise noise and especially DR again.

 

Noise: I think with RED MX and ALEXA no longer showing nasty noise beyond 400ASA and pushing the boundaries, we don't have to discuss it. You can push 5219 propably one stop to 1000ASA, while the ALEXA holds up nice files up to 2 stops beyond. You want ultra-low-light-photography (remember, 500/1000ASA and T1.3 is already great), digital is the way to go.

 

My conclusion: Highest IQ-possible with high-sharpness, clear images and barely any artifacts making it look "odd"? Easy: S35 with a high-quality 4k DI gives everything, versatility, robust workflow - it's even quite cheap when using a quick&dirty telecince or an optical print first. Modern 35mm has nothing to do with "soft" or "grainy" - it isn't even warm, cold, contrasty or flat - it's what you want it to be thanks to powerful digital post. And the images stand out against all the music videos and student films shot on RED...

 

All this will propably change in some years, but 35mm is the golden standard in most scenarios - it has many advantages and little disadvantages - no need to jump on the digital train until it's really ready. Let's see how well Deakins' work with ALEXA holds up on the big screen - maybe it's the first step into being more than just the "producers darlin'" because it's cheap...

 

Sources:

www.provideocoalition.com

www.arri.de/camera/tutorials/4k_systems_theory_basics_for_motion_picture_imaging.html

motion.kodak.com/motionloadedFiles/TI2647.pdf

Edited by georg lamshöft
Link to comment
Share on other sites

I do know that still images shot RAW with my Canon 5d look normal when processed in Adobe Camera RAW. The RED R3d files need some unusual tweaking to look "photographic" for lack of a better word.

 

Right after I graded the RED movie, I graded a project shot on the Sony f-35. Right away the images looked the way I imagined, though they have a bit more noise and are not as detailed as the RED images. But, color wise, they are much more "natural" looking. The Sony seemed to have about the same dynamic range as the RED MX, but does require a bit more light for best exposure.

 

Many camcorders have adequate color out of the camera, some Panasonic models and the higher end Sonys are in this category. They are generally ok in the midrange but when you approach overexposure they do produce false color and unnatural digital overexposure though, especially the Panasonics. The Sonys have better gamma curves. Overexposure aside, the Canon SLRs are pretty good also, there is evidently lots of design experience in these designs which makes sense considering the experience of the manufacturers and the years they have in the digital market.

 

The problem with RAW cameras is that RAW power is usually a substitute for spending more time in imaging design. The RAW signal itself is just photon counting on the 3 sensor types of pixel, all cameras are equally good if we ignore noise level and the infrared problems some manufactuerers allow to get higher sensitivitity. The sensor is basically irrelevant to the color quality of the output or overexposure characteristics. The only thing it affects is noise levels. The TONE itself is in the matrix and gamma curves, it's where manufacturers need to put more work. The sensor are technically wonderful these days.

 

The following images are all from the same Kodak sensor. The exposure level is slightly different but it does demonstrate what gamma is capable of:

 

vividrec709.jpg

 

The 1st and 3rd images are using a very agressive stack of "vivid negative"+"film print" model. The 2nd and 4th images are using a spec Rec709 gamma. It would be easy to blame the sensor for the differences, claiming one cannot achieve saturation and has bad overexposure, but it's the same sensor. Blame the designer instead:)

Link to comment
Share on other sites

 

All this will propably change in some years, but 35mm is the golden standard in most scenarios - it has many advantages and little disadvantages - no need to jump on the digital train until it's really ready. Let's see how well Deakins' work with ALEXA holds up on the big screen - maybe it's the first step into being more than just the "producers darlin'" because it's cheap...

 

Sources:

www.provideocoalition.com

www.arri.de/camera/tutorials/4k_systems_theory_basics_for_motion_picture_imaging.html

motion.kodak.com/motionloadedFiles/TI2647.pdf

 

Certainly Deakins', Richardson's, and Bay's work on the Alexa will prove very telling. I won't disagree with anything you wrote except to mention that while film's mtf may approach 80lp/mm ("true almost 4k") it slopes off quite a bit and so seems less sharp, whereas digital has superior micro-contrast, providing high mtf to extinction, and much less grain if also a bit less resolution. (Look at these comparisons: http://www.boeringa.demon.nl/menu_technic_ektar100_resolution.htm ) I remember my comparisons of 135 film scans and digital SLR files were very similar, the film had more resolution (approaching the 80lp/mm my scanner could provide) but it was "structurally" different and did not hold up when enlarged, really. Imo, it really is about aesthetics more than technical merit at this point, which is a big step for digital capture.

 

I think the "red look" is real and has to do with color rendering, not sure how, though. Even my 7d footage does not have that look; the skin tones are better. That said, the Red is a thousand times the camera and the newer footage with it (Pirates in particular) looks superb, technically and aesthetically.

Edited by M Joel Wauhkonen
Link to comment
Share on other sites

  • Premium Member

Tom, welcome back!

 

R,

Er, that post is about 8 months old.

Tom's now exclusively making his outrageous predictions over on Reduser, although there's scant mention of any sort of putting his money where his mouth is :lol:

 

Someone has let him fiddle with an Epic for a few hours, and I think he's too overcome to say much....

Link to comment
Share on other sites

  • Premium Member

won't disagree with anything you wrote except to mention that while film's mtf may approach 80lp/mm ("true almost 4k") it slopes off quite a bit and so seems less sharp, whereas digital has superior micro-contrast, providing high mtf to extinction

 

No, that is NOT ""true almost 4k"".

80 lp/mm means 80 black and 80 white lines per mm, or about 4000 lines over a 35mm frame width, that is correct.

 

However, to correctly resolve 4,000 lines you need AT LEAST 8,000 horizontal pixels, not 4,000.

This has been discussed and demonstrated many times here and other places, to little or no avail.

Electronic camera manufacturers have virtually from day #1, deliberately obfuscated the difference between optical "lines" and number of photosites on a camera.

They also misuse the term "resolution" itself, confusing nunmber of photosites with actual resolution.

 

 

At present, I don't think anybody ever scans at 8K resolution because there is no infrastructure that can usefully use that much data.

But unlike all digital systems, the data is still there for future use if required.

Whereas if you shoot something in say 3K x 1.7K (The "Down a Well" resolution of a RED One), it will forever be 3K x 1.7K.

 

The reality is, most cameras (and compression codecs) work far better with test patterns than they do with real-word images.

 

I have yet to see a resolution chart from RED that I thought was in any way technically meaningful.

They include responses that are over 30dB down, which for all practical purposes may as well not be there

 

 

If they can produce an 8K Epic, THEN they will have something that equals or at least approaches the useable resolution of film.

Link to comment
Share on other sites

80 line pairs per mm * two lines per line pair * 24.9mm per image width is just under 4000 lines per image width. So just under "true 4k." This is ignoring the fact that 20-30mtf or whatever 80lp/mm equates with is basically mush, and super mushy once you introduce a lens. Granted, according to nyquist theory you'd need 8000 pixels to resolve 4000 lines (2000 line pairs) without aliasing, sure, but no one filters that aggressively because such mushy detail doesn't induce visible aliasing. If this were a real issue, 4k film scans would alias severely and dSLRs would resolve 1/4 their stated resolution; neither is the case. And look at Fuji's 500T mtf charts. They're terrible. Most film isn't that great and it's been "good enough" for years. 80lp/mm does not actually happen on color film, except maybe in a lab.

 

Of course not of this matters at all. If you like the look of one thing, use it. From a technical perspective, I think it's a shame digital manufacturers omit MTF data for their sensors, but what's it matter, really? If it looks good it looks good. If it looks bad, it looks bad.

 

It's subjective, one person will tell you film resolves "8k" another will say far less (and this is a vistavision sized frame vs an APS-C sensor):

 

http://www.luminous-landscape.com/reviews/cameras/d30/d30_vs_film.shtml

 

Who cares so long as it looks good? I'd rather seen better skin tones on the red than all the Ks in the world.

Edited by M Joel Wauhkonen
Link to comment
Share on other sites

  • Premium Member

Can you blame me for being an optimist Keith, I only see the good in people.

 

R,

I was trying to somwhow massage up a reply that implied I thought you said "optometerist", which would hence enable me to go on make some cutting comments on certain people's visual acuity, but I fear it's too late in the day.

Link to comment
Share on other sites

I can see 1080 luma lines PPH on a synthetic 1080p source debayered with a good algorithm. The problem is real sensors with real lenses. We are dealing with 100lp/mm today.

 

In any case, I wouldn't say a 1080p source is 540 lines PPH, because I typically see more than 900 lines. No difference from a 3CCD/3CMOS source really in luma, because 3CCD/3CMOS is also limited by sensor MTF and lens MTF, the difference is that the debayer losses are substituted by the beam splitting prism losses. The 3CCD typically allows luma to alias instead of using an antialias filter, but OLPF only changes midrange MTF and can be compensated using digital post filtering.

 

You do not need 8000lines to resolve 4000lines (2000 line pairs). By sampling theory, you need half of that. You need 100hz sampling to resolve 50hz and 100 lines to resolve 50 line pairs = 100 lines. You do need 2x for resolving lines of chroma, but we are less sensitive to chroma and chroma is typically distributed with 4:2:0 formats so the demands are 1/2 of luma.

 

We usually agree that 1920x1080 HD with 5um pixels is sharper than 16mm film. Some HD networks do not even consider 16mm sharp enough for HD programming. Why don't we also agree that a 5um s35 academy/s35 sensor is sharper than 35mm academy/s35 film? From a lp/mm point of view it's exactly the same.

 

But the digital issues have nothing to do with resolution, the problem is the tonal response, the digital look. I don't see many people complaining HD is not as sharp as film. It's sharper than a 2K scan which is adequately sharp in my book.

 

Reason for editing: I had a few words disappear in random places in the post?!

Edited by Otis Grapsas
Link to comment
Share on other sites

whereas digital has superior micro-contrast, providing high mtf to extinction, and much less grain if also a bit less resolution. (Look at these comparisons: http://www.boeringa...._resolution.htm )

 

That's correct but we are talking about the world of cinematography, the OLPF has to be much stronger to rule out aliasing which is still accepted in the still photography world but more importantly, the frequencies we're comparing are usually much lower where modern film has very high MTF. A 24MP-DSLR will be compared to film at 80 cycles/mm while 1080p/2k on 35mm is 40 cycles/mm. Neither grain or MFT is really an issue on this level (at least when not using constantly underexposed 500ASA film-stock). The link is interesting because it shows behaviour at 40 cycles/mm but with worse film/scanner-combo and a ridiculous 200$-Minolta-lens which easily has 20-30% less contrast at 40 cycles/mm.

 

But I think we can agree that frequencies like 80 cycles/mm are difficult to handle in the processing-chain - but that's also the case with 35mm-style digital cameras. I think with 4k DLPs on large screens, 65mm makes more sense as an acquisition than ever.

 

We usually agree that 1920x1080 HD with 5um pixels is sharper than 16mm film. Some HD networks do not even consider 16mm sharp enough for HD programming.

Actually, with modern equipment (Vision2/3 and 3k scan) it's pretty close resolution-wise. But grain becomes an issue that has to be dealt with, especially for 3MBit/MPEG2-broadcast...

 

But the digital issues have nothing to do with resolution, the problem is the tonal response, the digital look

 

I agree but the "K"-argument will be discussed (and missunderstood) with the "5k-revolution" in the future as well, I fear. RED uses a reasonable sensor-design but doesn't want to give up the 4k-myth. Here is a comparison between 4k and HD output from 4k REDRAW: http://www.stopp.se/lab/videos/RED_4K_vs_HD.zip

Link to comment
Share on other sites

  • Premium Member

Gerog,

 

I don't get it. The 4K render is clearly sharper. And the HD render looks good, but it's downrezed from shooting the Red at 4K. If you don't think the Red is the sharpest digital camera out there, then please show an example of one that is as sharp or sharper.

 

Red doesn't say 4K gives 4K resolution. 4K is the sensor's (horizontal) pixel count. They say 4K gives about 3.2K luma resolution.

Link to comment
Share on other sites

The limit is obviously the lens when pixels get this small. I have done many tests with 5.5um pixels using 16mm lenses and some lenses are very soft even stopped down. The lenses are not really designed for this resolution. Going to 2um pixels wouldn't change the MTF. More chroma resolution but not sharper and a great loss on dynamic range and noise. In real shooting with DOF and motion, I doubt even 7um would look softer to the viewer.

 

We can show film and digital to be very sharp with tests, but is the lens setting and setup typical in filmmaking? Is the ideal aperture typical? Is a flat chart a typical subject? Are subjects as static as a test chart? If you put a highpass filter on a digital release you will rarely see anything pass the filter, unimportant detail on an occational wide angle setup perhaps for a couple of seconds. Real world vs theory. It would only be noticable on a direct comparison, one switching A/B images one on top of the other, probably. We sometimes forget that cinema is not static technical imaging, it's storytelling in motion. (storyshowing for the writers in the forum!).

 

Advancing the technology is good, but the limits of perfectly acceptable are lower than 720p IMHO and the audience does not really care if it's 4K or 8K. What they do care is the aesthetic of the image, the saturation abilities, the skin tone, the tonality. They do perceive those, even if it's downsampled to a DVD or to internet video and heavily compressed. You can touch the audience with low or high saturation and you can involve them with a good skin tone. The character will come to life, it will become something the audience can relate to.

 

If digital video wants to mature, it needs an aesthetic approach, not more pixels. The extra pixels always come at a cost in pixel quality, workflow, processing and archival anyway, nothing is free. If digital developers do not understand why a cinematographer might choose different stocks for different tonal results, film will always have a market.

 

I'm quite young, 35 at the moment, and my generation has shot very little film in motion. Even when we have, we never had enough time and exposure to the medium to experiment with different stocks. Many are shocked when they notice the tonal advantages of film but some are so immersed into generic digital imaging from their camcorder background, that they accept many of the digital artifacts and drawbacks as normal or even advantageous. Most of us are not aware of the tonal differences of different stocks. This is a cinematography forum. Where are the stock comparison threads in subjects that involve skin tone?

 

All this supports the market for high resolution digital camcorders and establishes resolution as a main direction for future improvement. What my generation generaly use to compare filmmaking equipment is resolution and noise, there is no aesthetical decision making whatsoever. I have seen countless comparison of A (wonderful skin tone with a little noise) vs B(terrible skin tone with a flat clean and sharp computer graphics look) and 90% of the times, the second image is what the majority prefer. I'm not talking about the audience, the audience would choose the better skin tone, I'm talking about us, the half educated videographers that read technical articles, do technical comparisons and write Director of Photography on our business cards.

 

I don't believe this will change for many years. The market is shaped by the majority and going after aesthetics in digital imaging is a very small niche the majority does not care about. Not large enough to support a business really. It's like trying to operate a business on high end tube power amplifiers in a world of integrated video receivers with multichannel amplifiers and a great selection of surround decoders and DSP.

 

Another problem I see is that with RAW bayer, digital image development becomes very generic. We have already seen this in still photography. While the manufacturers pack some good aesthetic engineering into in camera production of JPEG and their own RAW development apps, Adobe DNG will make sure the differences will be eliminated at some point. Everything will happily look the same with generic processing designed to match the generic medium quality still camera and hide its flaws. With Cinema DNG we will see the same phenomenon in the digital cinema market.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Forum Sponsors

BOKEH RENTALS

Film Gears

Metropolis Post

New Pro Video - New and Used Equipment

Visual Products

Gamma Ray Digital Inc

Broadcast Solutions Inc

CineLab

CINELEASE

Cinematography Books and Gear



×
×
  • Create New...