Digital Media | Spotlight: Photography | Inside Lightroom | Blogs
Digital Photography: Where do we go from here?
In an effort to keep this blog going I thought I'd submit another excerpt from my newsletters. This article was originally published in the Fall 2007 issue of my newsletter and that issue is available online here. Basically this is my list of what we really need in new DSLR and Medium format digital cameras.
I originally thought up this article about a month ago just before Nikon and Canon announced some of the most ground breaking cameras to ever hit the market in the fall of 2007. In light of these new digital SLR camera models I thought I'd still go ahead and give my opinions on "Where do we go from here?" and add my comments relating to the latest cameras. [Please note: I have updated parts of this article for this blog post.]
I have written before about maximum resolution in 35mm type digital SLR cameras so I won't go into that here. Suffice it to say that I think we have hit the wall or are pretty much right there in terms of how many megapixles are practical and possible in DSLRs. Canon's recent announcement of the 1Ds Mark III with its maximum resolution of 21 MP is going to push Canon's lenses very hard. And if the rumors hold and Nikon releases a 20+ MP camera then we'll have come very close to the limits of 35mm optics since the lenses are the limiting factor - or rather the diameter of the lenses.
So without further ado here is the list of what I feel we really need in new digital cameras, for both 35mm and medium format:
Higher Bit Depth: The first thing that I have had on my wish list for a long time is higher bit depth. Currently most digital cameras (35mm) operate at 12 bits (4,096 colors per channel). While it works quite well, highlight blooming is a common issue and transitions between highlights and shadows could be better. Hence, I am not surprised to see both Nikon and Canon introduce 14 bit cameras and this is a huge improvement in image quality that is reason enough to upgrade even if there were no other improvements in the camera.
One of the major reasons for a higher bit depth besides the issues above is that it also gives you a lot more headroom when tweaking images in photoshop. Now, I don't go to extremes in general. I am one of those photographers that likes to get it right in the camera - but every once in a while when I pick up the artistic license it is nice to go a little crazy with an image and not have severe noise or pixelation occur.
On a side note, medium format backs have been 14 bit for a long time (even though most advertise their backs as 16 bit). The only medium format back that is truly 16 bit are the Leaf backs according to my sources. True 16 bit is fantastic and in the future, we can only hope that the camera manufacturers will aim for true 16 or even 24 bit capture.
Better Lenses: The optics are currently the limiting factor in terms of resolution for 35mm based digital cameras. This has more to do with physics than it does our ability to build incredible lenses. Nonetheless, better made optics and optics made specifically for digital capture would help eliminate some of the issues that commonly appear in digital images such as chromatic aberration, vignetting, loss of sharpness and overall image quality. This is especially true with wide angle lenses and both Nikon and Canon can improve their optics in this genre.
Nikon has already started building digital specific lenses (14-24mm f/2.8 and the 24-70mm f/2.8) and is shipping these along with the new D3. On the Canon front, especially with their new 21 MP camera, their optics need a big improvement if they hope to keep up with that sensor. At the moment, I would venture a guess that most of their lenses don't resolve as much detail as that 21 MP sensor. Even with the 16 MP 1Ds Mark II, Canon only had a few lenses that could match the resolution of the sensor - I haven't done any testing on this but from what I have heard from Canon shooters it seems to be true. And with my Nikon D300 and D700, I have found that it pushes my lenses very hard - only the best lenses can really resolve what those sensors can produce.
As a side note with medium format digital capture the optics have not yet become a limiting factor because of the larger diameter lens mount and we haven't yet reached a resolution that really taxes those lenses but we are getting there fast with the newer 65 MP cameras. Even so Hasselblad has designed specific digital lenses, mostly wide angle options, for the latest H3D.
Wider Dynamic Range: As a pro photographer who often shoots in challenging lighting I am always wanting greater exposure latitude, especially in the highlight areas. Digital has significantly more detail in the shadows than slide film ever did and we can't complain there at all. It is only in the highlights that I wish we could improve the dynamic range of these cameras. Part of this will be helped by higher bit depth (as discussed already) and part of this will have to be addressed via firmware as it has been with the latest Canon 1D Mark III. I applaud Canon for including what they call "Highlight Protection" in their latest cameras as this helps retain highlight detail. I have heard that Nikon has incorporated something similar into the Nikon D3, D300 and D700 but it remains to be seen how effective it is in everyday usage. This is an area where overall image quality can be drastically improved. We may not always need extreme exposure latitude but it gives us many more options in the final processing and obviously in contrasty and difficult lighting conditions.
Live and accurate Histogram in the Viewfinder: It might be considered a bold move by some, but if I had my way I would take the standard metering nomenclature out of the camera's viewfinder and replace it with an accurate live histogram. Why? Because for metering these days the camera's meter is only the starting point for a good digital exposure. Using the histogram to judge exposure is a much more accurate and better system then setting your camera in manual mode and just clocking the exposure where it tells you. This would allow us to work faster and really lock in the exposure without having to look down at the back of the camera and check out the histograms. Even better if they could have a "heads up display" on the screen in the viewfinder that splits out the red, green and blue histograms after each shot that would be incredible. And last but not least - if we could have accurate histograms from the actual raw images instead of histograms built from a jpeg rendering that would be HUGE!
Lose the 35mm format frame: This might just be a personal preference, but I am not a huge fan of the overly rectangular 35mm frame ratio (2:3). I would much prefer to see a 6x7 or 6x45 type framing ratio since the 35mm frame has to be cropped quite a bit for the normal magazine page. Nikon has caught onto this idea by offering different crop factors with the new D3. I think another reason I'd push for a less rectangular sensor is that there is a lot of wasted space in the image circle with the standard 35mm frame. A 6:7 framing ratio makes better use of the lenses image circle and would likely have less vignetting as well.
More Accurate and Reliable Autofocus: With the advent of high resolution digital SLRs we have seen that the autofocus of the film days wasn't all that accurate. And it still proves to be a hot topic. Witness the big autofocus issues with the Canon 1D Mark III. I know of several sports and wildlife photographers that feel burned by Canon because the autofocus is less accurate than their previous camera. This is a quality control issue I think and for companies like Canon to release a pro camera with major autofocus issues is pretty rough, especially when buyers plunk down $5,000 and expect it to work! In my own experience, I have to send in my Nikon D2x about once a year to have the autofocus re-aligned because a few outer focus point start to go off kilter. With my newer D300 and D700 the AF is much better and having the fine-tuning custom setting takes care of most problems.
Lower noise at High ISO's: This has been an ongoing issue for Nikon shooters, not so much for Canon shooters. But now with the release of the Nikon D3, D300 and D700 we have a whole new era in low light shooting possibilities. Kudos to both Nikon and Canon for working on this. I have been shooting at up to ISO 3200 with my my D700 and the images are incredibly clean. This new feature has already opened up entire new worlds of photography. And I have a feeling it can get even better.
Faster framing rates for 12+ MP cameras: Canon just announced a 10 MP camera that shoots at 10 fps and Nikon's D3 has specs of shooting at 9 fps at 12 MP so it seems like great strides have been made in this area as well. As the technology continues to grow I'm sure we'll see even faster framing rates on even higher megapixel cameras. The new D700 also shoots at 8 fps with the battery grip and this is a great combo for any photographer.
Better VR & IS lenses: It would be very nice to see the Vibration Reduction (aka Image Stabilization) improved as well to help counteract the high resolution these cameras can capture. When you increase the number of pixels on a sensor you also increase the sensitivity to camera shake and at 21 MP with the latest Canon that will become a huge factor. Don't get me wrong - the current crop of VR and IS is already very good but I hope they keep working on it so it can keep up with the new high resolution cameras.
Smaller and Lighter Cameras: Last but not least, do these things have to weigh four or five pounds with a lens on them? Hopefully smaller, lighter, more ergonomic bodies and lenses are coming as well that won't sacrifice durability.
A Complete Workflow
As this is a Lightroom blog I also want to promote a complete workflow. My124-page updated and revised workflow e-book using Lightroom 2.0 includes information on all the new upgrades to Lightroom and how I integrate Lightroom into a complete workflow from camera to Photoshop. If you'd like to check it out click here.
Also, I have to Thank Ken Milburn who wrote a glowing review of the ebook right here on the Inside Lightroom blog. You can check out his review of the ebook here.
That's it for this week.
Adios, Michael Clark
If you are going to go, go big! That's quite a list.
For me, higher dynamic range and fuller colour representations are the two most important things I'd like to see improved before digital images lose the "digital" part.
Of course, with things as they are it is the eye behind the lens that makes the largest different. Taking digital photography beyond point and shoot and into the domain of something we might call art still feels like the domain of some few masters.
Compared to film photography, I feel like (semi-) pro digital equipment still holds me back. I don't have the raw talent nor the obsessive attention to every detail that makes the difference between a good shot and an excellent one.
Most of my near excellence is due to luck, and a little more forgiving media would help me with that. Going back to fundamentals of exposure and composition is always worth the time, and I try to practice this. But I do notice that sometimes the information captured from the scene just isn't caught at all, or poorly, in terms of tone, colour and dynamism.
A number of factors suggest that two of the items on your list are not yet close to being actual bottlenecks to image quality.
First, on the issue of better lenses, I'm not saying that better lenses aren't welcome. However, the evidence is that sensor resolution is not yet high enough to capture the full information available from many existing lenses. The resolving ability of a lens needs to be on the order of at least 5x worse than that of the sensor before the combined system resolution of the two fails to improve with further sensor resolution. The fact that many lens product tack-sharp images at, say, the 12Mpixel point says that we're not even close to seeing the rolled-off but still significant portion of the MTF curve of the many of the lenses we're using.
The second bit of evidence is that there are differences in apparent sharpness between cameras with different strengths of AA filters. Since Canon cameras tend to be conservative in this regard, it's likely that the high-frequency content of the image that a lens is projecting is being limited still as much by the AA filters as by the lens. Since the strength of the AA filter can weaken as sensor resolution improves, an apparent improvement in lens performance may follow as well.
On the issue of bit depth, many studies have shown that even at low ISO, the signal-to-noise ratio in the tone values most likely to benefit from higher bit depth (that is, to posterize if bit depth is too low and/or post-processing corrections too severe), the shadow tones, is far too high to actually benefit from greater than 12 bits. Until read noise is reduced, more bit depth buys nothing.
The argument that higher bit depth provides processing headroom is equivalent to saying that buying a TV in a larger box allows more flexibility in positioning it in your bedroom. In reality, you'd take the TV out of the box and position it freely within the space of the room. When capturing data at the lowest bit depth needed to fully represent the signal-to-noise of the signal, the in-camera or raw processor can still pad the resulting data with zeros (or even random bits) in order to turn the data into a higher-bit depth operand for further processing. That is, you'd take the data outof its 12-bit box and put it into a 16-bit one before applying post-processing steps. The end result is exactly the same. Capturing the data at a higher bit depth than required adds precision but not accuracy and the extra bit are typically just random white noise.
I do agree with most of your other points, however.
Michael,
I have been to several ASMP events in New Mexico, the most recent one was your lightroom talk in Albuquerque. I have been an assistant to Robert Reck and you would recognize me as the guy who kept asking you all those questions about lightroom.
I have given a couple of responses to a few blogs about what I think / like / desire to happen in the ever changing world of DigitaL Acquisition.
Here is my latest I posted to Stu over at the prolost blog.
Stu,
I apologize in advance for the long list here of rants.
The new paradigm:
All cameras must be out of the box usable by one person and have all capabilities of said camera usable by one person, the "one camera one person" approach not 5 people. Sorry 600. They must support lenses in field with total sensor contact. no lens disabling features allowed.
Same look and feel of regular DSLR, with a 4.5 or larger HD capable live view touchscreen . Helps get rid of a lot of buttons. Sorry shoulder mount guys but this is first a still camera, not some Ikegami based form factor.
Stop 35 mm sensors, go to 48 mm or larger for acquisition sensor.
Dual sensors, one large for raw/open dng acquisition, one for viewer screen and / or output to hd/hdmi output. Obviously some mirror / prism optics engineering here.
Total support for raw at all frame rates, which sort of makes NLEs possibly EOL'ed. Especially if Lightroom /Aperture / Photo-Mechanic start to support time lines.
The ability to have a real easy up-loadable parameter settings. So I can have my Lightroom preset be embedded in the viewer sensor parameters. Then my clients can have a real time near approximation of a final look if wanted. The paint box approach to looks on a still camera. From off the shelf software.
Absolutely no snap in or plug in hardware modules except for battery and maybe memory. Which leads to a all weather camera. Nikons are noted for their "takes a licking and keeps on ticking". As a former USAF photographer I can attest to that.
If audio is needed, allow non manufacturer connectivity. Open standard, audio is not a real need or make or break here. In the highest capture form, you should be capturing audio to a separate means. Maybe said camera also has a separate audio module that is either wireless or a small tethered device. Again no hardware snap in modules. Plus a time code generator or genlock thing.
"Where I look " focusing, With the hi end Nikons and Canons having multi point focusing systems this would be a real nice feature to have. Canon does this, called the Eye Controlled Focus.
The ability to have my flash usable during HD/ High frame rate capture, pocket wizards stuff like that with the ability to program my flash modules as needed or allow the camera to do its TTL thing. And have color temperature settings changeable on the fly.
Now to go into "Out-there land".
If the next macintoshes coming out are gestural based touchscreens, allow the user to be able to change lens or camera or whatever settings on the fly with real time viewer feedback systems. Tethered or wireless, I prefer tethered for reasons of reliability.
Have a new "Open Platform Stability Standard" which allows the creation of "Intelligent Monopods, Tripods, Steadi cam," systems so you can offload some of the processes or power/ memory requirements to the stability platform. The idea of "one person--one camera" to me makes this a nice addition. Though I can see someone like Lars von Trier doing a 100 camera setup ;-)
Finally:
If nikon is true to their word they are making probably ± 90,000 D90 cameras a month, opinion based on this link:
http://news.cnet.com/8301-10784_3-6168993-7.html.
Therefore, anyone trying to keep this camera above $3000 for body alone would be out of business soon.
Thats my dream camera, for now at least. Oh one more thing, must have the hi end canon/ nikon iso range...
Hi, The biggest advantage of digital photography over traditional film include: Instant review of pictures, with no wait for the film to be developed: if there's a problem with a picture, the photographer can immediately correct the problem and take another picture.
Regarding the live histogram - I set my Fuji S9100 to show the live histogram. Fuji says it's not quite as accurate as the one that comes up with the already-shot image review but it's a wonderful pre-shot tool that has made a world of difference in my results.