Camera industry in the dumpster - article

[bcooter]

The Stunt Man, a film that did nothing at the box office,

I saw stuntman years after it's release.    Obviously the clothes and pace were dated, but the story and in particular Peter O'toole's talent were off the scale.

I heard through the crew/studio/industry/BS rumor mill that the studio was pissed and I think held it back, finally opening it in Seattle.  It built a small following then they yanked it, but then there was no netflix, or twitter feeds.

There is a film about the film

http://www.imdb.com/media/rm2048498688/tt0238606?ref_=tt_ov_i

which details it.

Usually though shows about the industry fail.   A few make it but only if the distributors are 100% behind it and nobody gets worried.

Remember studio heads have good jobs . . . they don't want to lose them.

IMO

BC

[bcooter]

Not that I have any idea of motion but are these alternatives ?

http://www8.hp.com/us/en/campaigns/workstations/z820.html
http://blogs.windows.com/windows/b/extremewindows/archive/2012/11/13/video-editing-supercomputer-z820-red-edition.aspx

Yes, except it's just changing complete systems is tough.  Especially with multiple places, and traveling.

Then there is the problem that you can't make a pro res file or any quicktime in a windows box.  

QT may be leaving, but today it's the delivery standard, at least for my clients.

Proprietary to make money I understand, such as lens mounts (though I don't like it).

Proprietary just to make a statement like the new Mac Pro is an insult to the professionals that propped Apple up in the bad times.

All the talk today is 4k.     Yes you can edit 4k in proxy's or dumb down the NLE to play essentially a 1080 proxy, but work in real 4k and your going to glitch.  

This is where the camera makers aren't communicating with the software companies and the computer makers.

You can now shoot 4k on the cheap, but editing and delivery is far from simple, same with color and compositing.

No reason on this planet that Apple builds a new box that doesn't have a way to put 16 lane pcI cards to it, either onboard or off.

No reason that Apple tosses out fw800, at least leave one powered plug.   I mean it's hubris to think that professionals that have hundred of stored terabytes are all going to dedicated two months and a team to transfer everything to thunderbolt, especially since thunderbolt is not the most secure connection.

IMO

BC

[Ray]

You must have used a very old Canon like e.g. the 5D. All Canons since ca. 2007 has an AF-ON button,

Quite right. My only full-frame Canon DSLR was the 5D. I almost upgraded to a 5DII but was seduced by the impressive quality of the Nikkor 14-24/F2.8 lens, and the impressive DR of the latest Nikon DSLRs.

The disadvantage I found with the 5D, when using auto exposure bracketing to capture an optimal exposure, was that sometimes the best exposure for clean shadows was too slow for a sharp image.

Auto ISO bracketing with fixed aperture and shutter speed would have been a better option, especially considering how much the shadows are improved with increasing ISO on all Canon cameras. The greater ISO-less nature of the Nikon cameras lends itself to greater flexibility in this regard. Underexposure at a low ISO may be only very marginally worse, with regard to shadow noise, than the same exposure at a high ISO.

[Isaac]

Telecaster, that is exactly what any competent product management team does. They do not share the database, however, so in general you don't know about it.

Wondrous!

Might it even be possible they actually know how much profit they make on each of their cameras and lenses!

Of course, our unlimited speculation completely depends on not knowing any of the information that people on-the-inside actually have available to justify their decisions ;-)

[Telecaster]

I saw stuntman years after it's release. Obviously the clothes and pace were dated, but the story and in particular Peter O'toole's talent were off the scale.

I heard through the crew/studio/industry/BS rumor mill that the studio was pissed and I think held it back, finally opening it in Seattle. It built a small following then they yanked it, but then there was no netflix, or twitter feeds.

I guess I'm among the few folks who saw The Stunt Man during its original brief release. It played for maybe a week at one theater in the Detroit area. My friend Dale & I went to an early evening show...there were less than a dozen other people there. I wanted to see it mainly 'cuz I thought Barbara Hershey was cute.    Afterwards we both thought it was one of the best films we'd ever seen. It doesn't hold up to that standard now—as you note it's of its time, and its type of twist ending has since been done to death—but I watched it again not long after Peter O'Toole died and was still impressed. A lot of my favorite films are ones with huge ambition that may not quite reach what they were going for but still take you on a real good (& sometimes wild) ride. The opening sequence of The Stunt Man is a classic IMO.

-Dave-

[Telecaster]

The discussion of optimal exposure reminds me of audio engineers who feel they still have to push recording levels just short of 0dB on the digital scale, even though with 24-bit recording you can hold peaks to -12 or even -15dB for the sake of headroom and still preserve high fidelity while maintaining an inaudible noise floor. We don't have quite that degree of latitude with electronic photography, but with the cameras I'm using now fussing over exposure to such an extent gains me nothing. The quality of shadow gradation I'm seeing is very impressive even exposing for optimal JPEG use. It's a surplus of data actually, and thus I no longer have to worry about it. This is good!

-Dave-

[Hans Kruse]

The discussion of optimal exposure reminds me of audio engineers who feel they still have to push recording levels just short of 0dB on the digital scale, even though with 24-bit recording you can hold peaks to -12 or even -15dB for the sake of headroom and still preserve high fidelity while maintaining an inaudible noise floor. We don't have quite that degree of latitude with electronic photography, but with the cameras I'm using now fussing over exposure to such an extent gains me nothing. The quality of shadow gradation I'm seeing is very impressive even exposing for optimal JPEG use. It's a surplus of data actually, and thus I no longer have to worry about it. This is good!

-Dave-

Dave, if it is good for you and the way you photograph and your requirements, then that's all good. The point I was making was not about this.

[Ben Rubinstein]

The discussion of optimal exposure reminds me of audio engineers who feel they still have to push recording levels just short of 0dB on the digital scale, even though with 24-bit recording you can hold peaks to -12 or even -15dB for the sake of headroom and still preserve high fidelity while maintaining an inaudible noise floor. We don't have quite that degree of latitude with electronic photography, but with the cameras I'm using now fussing over exposure to such an extent gains me nothing. The quality of shadow gradation I'm seeing is very impressive even exposing for optimal JPEG use. It's a surplus of data actually, and thus I no longer have to worry about it. This is good!

-Dave-

Thing is that with audio you just listen to it. A raw image is just the very beginning of the processing pipeline. The information is there so that we can get to that final low fi jpg.

[Hans Kruse]

Well, I must admit I don't spend much time fussing with exposure in the field; I take a look to see whether I clipped something and usually make minor exposure adjustments in raw conversion, but other than that, I have a "knack" for nailing exposure from years of shooting chromes, which are quite unforgiving in both shades and highlights. Compared to that, digital is so easy to get good results with I'm just not motivated to geek it out.
Also, most of the stuff I shoot doesn't lend itself well to ETTR principle; if I shoot handheld in variable lighting conditions, it's difficult to nail it as it is, but if I add another variable it spirals out of control. For instance, this:


Go even the tiniest bit to the right and it's ruined. It's so vital to guard the highlights, nothing else matters, especially with a modern large-sensor camera that renders blacks so velvety smooth.

ETTR means that you exposure maximum without blowing out any highlight. The way I shoot I preserve highlights and I agree, of course, that any essential highlights should not be blown out. If you have a low DR scene you can underexposure by one stop and no problem, but if you have a high DR scene this is not good if you want to see details also in the shadows. Therefore I have adopted my approach which also lends itself to do an HDR merge if needed. So all cases are covered. The downside is many more exposures. The upside is no wasted time or attention to non essential technical details when shooting. In many cases there is time enough but when the light and weather really plays in the landscapes then one need to maximize the attention on the shooting, compositions, moving around etc. In such situations errors are made by even the most experienced shooters Believe me I have been there ....

[hjulenissen]

The discussion of optimal exposure reminds me of audio engineers who feel they still have to push recording levels just short of 0dB on the digital scale, even though with 24-bit recording you can hold peaks to -12 or even -15dB for the sake of headroom and still preserve high fidelity while maintaining an inaudible noise floor. We don't have quite that degree of latitude with electronic photography, but with the cameras I'm using now fussing over exposure to such an extent gains me nothing. The quality of shadow gradation I'm seeing is very impressive even exposing for optimal JPEG use. It's a surplus of data actually, and thus I no longer have to worry about it. This is good!

-Dave-

Imagine the joy of those audio engineers if the manufacturers told them "sorry, you won't be able to add your required N dB of headroom based on metering of raw inputs. Rather, we'll do some dynamic range compression and mp3 encoding of the audio first, then provide you with level estimates from that output".

-h

[LKaven]

Imagine the joy of those audio engineers if the manufacturers told them "sorry, you won't be able to add your required N dB of headroom based on metering of raw inputs. Rather, we'll do some dynamic range compression and mp3 encoding of the audio first, then provide you with level estimates from that output".

Good one! 

Really, as an audio engineer, I would not settle for -12 to -15dBFS peaks.  Those are more like ENG or film dialog levels.  I really do prefer to get it up at least to -6dBFS or better.  A little fast compression and limiting on the way in is not a bad thing either.  Usually it's only a very few things that blow your levels -- big tom-tom hits, etc. 

And yes, you can really hear the benefits of preserving bits in audio.  First of all, there are no real 24 bit converters.  The most you can meaningfully get is 20-21.  But that's /a lot/ better than 16.  You can hear it in the reverb tails.  The more processing you do (we work in a 64-bit float space) the more you can appreciate how those extra bits help. 

[jjj]

I guess I'm among the few folks who saw The Stunt Man during its original brief release. It played for maybe a week at one theater in the Detroit area. My friend Dale & I went to an early evening show...there were less than a dozen other people there. I wanted to see it mainly 'cuz I thought Barbara Hershey was cute.     Afterwards we both thought it was one of the best films we'd ever seen. It doesn't hold up to that standard now—as you note it's of its time, and its type of twist ending has since been done to death—but I watched it again not long after Peter O'Toole died and was still impressed. A lot of my favorite films are ones with huge ambition that may not quite reach what they were going for but still take you on a real good (& sometimes wild) ride. The opening sequence of The Stunt Man is a classic IMO.

I saw it at a preview showing as I was reviewing the film for the student newspaper I worked for. It's one of my favourite films too. So I gave it a very positive review, bar the dip in the middle for the romance scene, which made for a lurch in the momentum of the film. I'm pretty sure have the soundtrack for it in my vinyl collection too. Considering the film got nominated for 3 big Oscars, director, best actor and screenplay, it's strange it got such a limited release. But then studio heads can be complete dicks.
Sadly I doubt the DVD release will have gained it many new fans as it will have suffered from not being the fresh and original film it was back then, for the reasons you mention.

[dturina]

ETTR means that you exposure maximum without blowing out any highlight. The way I shoot I preserve highlights and I agree, of course, that any essential highlights should not be blown out. If you have a low DR scene you can underexposure by one stop and no problem, but if you have a high DR scene this is not good if you want to see details also in the shadows. Therefore I have adopted my approach which also lends itself to do an HDR merge if needed. So all cases are covered. The downside is many more exposures. The upside is no wasted time or attention to non essential technical details when shooting. In many cases there is time enough but when the light and weather really plays in the landscapes then one need to maximize the attention on the shooting, compositions, moving around etc. In such situations errors are made by even the most experienced shooters Believe me I have been there ....

Yes, I agree with you that if you really need information in the shadows, it might be a good idea to blend several exposures in HDR software, but honestly, I never had good results with that. To me, it looks like geeking it out, like, let's get all the information we can, but my experience is that photography is as much about letting information go, as it is in obtaining it. You need to discard stuff. This desire to include everything, to mourn every white and black pixel as a tragic loss due to clipping which would be better recovered, I don't know, I can't really adopt that attitude. For me, it's like this: I know what I want to get and if I'm getting it, I'm happy. To me, it's getting the blacks to look clean and smooth, defined by the subject and not by sensor readout noise or banding. If the blacks look good, I'm quite satisfied to leave them black, I have no great need to push 2ev of shadow detail out of it and make blacks look muddy brown. Actually, the point where I fell in love with a large-sensor camera is when I saw a picture of deep dark rock on cascading waterfalls, rock that was perfectly defined and yet deep black, and I understood that I couldn't make that shot with either C41 or E6 film, or my small-sensor digital. The silde has no color definition in the shades, the negative would define it as an empty hazy blob and the small sensor would also define it as brown due to noise. But only a large-sensor digital would define it as pure, 3d volume of darkness.
ETTR technique assumes there's a problem with shadows when shooting digital, but I can't confirm that, not with low-ISO shots with modern large sensor cameras. In fact, I find the opposite to be true; the shadows are defined in the most perfect tonality, it's the greatest asset of digital photography compared to film. Also, Michael's explanation which says that the left side of the histogram is defined by a very limited binary number-space sounds flawed; in my understanding, the sensor element uses photoelectric effect to charge a capacitor with photons, and at readout the ADC converts the charge of the capacitor, from 0 V to MAX V, into a binary number that defines luminance. Why LSB of that number would carry more information than MSB, I have no idea. The information is limited by photon count, by the readout noise that could corrupt the information in pixels adjacent to the readout conduits, by the ADC noise, EM noise from the PCB and similar things, but once the ADC has delivered its number, all bits are equal. Sure, if a pixel is so dark it has a luminance value of 3, there isn't much between that and zero, only 2 bits of data to work with. But if you're working in 8-bit space and you're working with pixel brightness of 253 to 255, you're still working with those same two least-significant bits to the left, because everything else is set to 1. So the entire logic that refers to the limited number space at the left side of the histogram, compared to the right, is flawed.

[bjanes]

So the entire logic that refers to the limited number space at the left side of the histogram, compared to the right, is flawed.

That observation has been made some time ago by Emil. It is not the number of "levels" but rather the signal to noise ratio that is the proper rationale for ETTR. One should not clip highlights by overzealous ETTR, but even though the histograms on current cameras are flawed, you are safe if the RGB histograms show no clipping, since these histograms are conservative.

Bill

[hjulenissen]

ETTR technique assumes there's a problem with shadows when shooting digital, but I can't confirm that, not with low-ISO shots with modern large sensor cameras. In fact, I find the opposite to be true; the shadows are defined in the most perfect tonality, it's the greatest asset of digital photography compared to film.

I don't get your reasoning here.

Any exposing assumes that there is a "problem" with your camera (combined with light itself): it has limited DR. If DR was unlimited, there would be no need to think about exposure, just set exposure time and aperture from purely aesthetical goals wrgt DOF, motion blur etc.

ETTR is simply an advanced name for something that digital audio engineers have practiced for a long time: keep signal levels high, but avoid clipping. The two goals conflict, and in practice you should find some compromise. The better feedback you receive, the more aggressive you can do that compromise.

Of course, if your gear has far more DR than your requirements, you can relax the compromise (and rather focus on other aspects of photography).

-h

[jjj]

To me,  it looks like geeking it out, like, let's get all the information we can, but my experience is that photography is as much about letting information go, as it is in obtaining it. You need to discard stuff. This desire to include everything, to mourn every white and black pixel as a tragic loss due to clipping which would be better recovered, I don't know, I can't really adopt that attitude. For me, it's like this: I know what I want to get and if I'm getting it, I'm happy. To me, it's getting the blacks to look clean and smooth, defined by the subject and not by sensor readout noise or banding. If the blacks look good, I'm quite satisfied to leave them black, I have no great need to push 2ev of shadow detail out of it and make blacks look muddy brown.

My view exactly.
If it looks good it is good. I don't care if the histogram is combed or if I've not captured every pixel I could when the end result is what I require. With film I sometimes used Kodak recording film, not the best quality film in the first place and then pushed it in a speed developer. This 'degraded' quality further, but I really liked the end result, so that was fine.  


ETTR has its place too, if you are shooting an image where the histogram data is clustered down the dark end and if you ETTR so that the lump of data moves to the right and doesn't get clipped then in theory you get a more malleable image. Whether it 'looks' better is purely an aesthetic choice. It also depends on the sort of work you are doing. Landscapes and other other similar considered work where you have time to tweak, ETTR is an option. If you shoot more spontaneous work, then shooting 'a bit dark' as I heard it described the other day [by a very well paid wedding photographer] helps to preserve highlight from being blown. This is more important than the potential gains of ETTR.

[jjj]

ETTR is simply an advanced name for something that digital audio engineers have practiced for a long time: keep signal levels high, but avoid clipping. The two goals conflict, and in practice you should find some compromise. The better feedback you receive, the more aggressive you can do that compromise.

Signal levels were kept high to stay above the noise floor, a practice needed with older audio equipment that isn't really analogous to what photographers do.
Poor quality sound recording, simply sound bad. However poorer quality photographs can look great as it can be deliberately done as stylistic thing, so not necessarily a compromise at all. This is why a well shot film where the sound is off is perceived as a badly made film. Whereas a film with deliberately lo-fi/poor quality camera work and high quality sound will be seen as well made.

[hjulenissen]

Signal levels were kept high to stay above the noise floor, a practice needed with older audio equipment that isn't really analogous to what photographers do.

Signal levels and noise levels are topics of interest for audio and image people alike. "why are my ISO50000 shots so noisy??"

Poor quality sound recording, simply sound bad. However poorer quality photographs can look great as it can be deliberately done as stylistic thing, so not necessarily a compromise at all.

Musicians have probably used "lowfi" recording as a stylistic thing for nearly as many decades as they have been able to record at all. The most prominent example is perhaps overdriven tube amplifiers and crappy 12" speakers used as sound shapers for electric guitars. A more recent thing is the popularity of early 8/12-bit samplers and drum machines, vinyl record simulation, etc.

The question is not if "poor quality" can be used for artistic purpose (of course it can). The question is if stribing for "high quality" can be a worthwhile thing to do.

-h

[dturina]

That observation has been made some time ago by Emil. It is not the number of "levels" but rather the signal to noise ratio that is the proper rationale for ETTR. One should not clip highlights by overzealous ETTR, but even though the histograms on current cameras are flawed, you are safe if the RGB histograms show no clipping, since these histograms are conservative.

Bill

I agree with the S/N rationale, that's the part I have no problem with, with the disclaimer that modern sensors have very clean shadows at base ISO and the problem there isn't so obvious as to warrant looking for a solution.

[dturina]

I don't get your reasoning here.

...
Of course, if your gear has far more DR than your requirements, you can relax the compromise (and rather focus on other aspects of photography).

Exactly. Not only do I have more DR now with digital than I remember having with C41 film, but also the noise is less than I remember having anywhere. In fact, ISO 800 digital looks about the same as scanned Sensia 100. Digital is so insanely good for so many years already, I actually feel ungrateful if I find myself nitpicking.