Canadian TV, Computing and Home Theatre Forums banner

1 - 12 of 12 Posts

·
Registered
Joined
·
521 Posts
Discussion Starter #1
The Megapixel war continues. The new Nikon and Pentax DSLRs are now boasting 16 MP sensors - possibly the same sensor.

Meanwhile the newest medium format sensor now has 80 MP and claims 12-stops of dynamic range.

The most recent rumour regarding the upcoming Canon 1Ds IV is that it will have 29 MP. What is more interesting, though, is that some believe that a new 5D Mark III will be released before the 1Ds and will likely contain 28 MP.

If a new 5D appears, it will likely be driven by a new processor, a Digic V. And then if the 1Ds is released later, then it should contain the Digic V as well, meaning that this model will skip the IV processor altogether.

Out of curiousity I looked at the Specifications for the sensor size for the 7D and the 5D II and calculated how many Megapixels the 5D would have if it were packed as densely as is the 7D. And the answer is: over 46 MP.

So a 28 MP 5D III is not unreasonable. Especially given that noise reduction (or the ability to separate noise from signal) seems to gain about 1 to 2 stops each generation. That is, it is not unlikely that such a model would boast 28 MP, but be able to shoot up to ISO 51,200. And if they would put a dual processor into it and have it fire at 5fps, I would be satisfied.

The upcoming 1Ds is still a puzzle. Ditto for Nikon's flagship 24.5 MP model replacement. It seems like the two companies are trying to get the other to release its new model first. Expectations of a breakthrough camera are high. And Sony is pushing both of these companies. Sony does make the 24.5 MP sensor used by Nikon, yet the Nikon takes much better images than does the Alpha 900. So Sony knows that by improving processing of its image data, it can become a viable high-end competitor.

One of the advantages of a large sensor for me lies in cropping. I do some wildlife photography and often the subject is far away, even with a big telephoto lens. So I crop. However, if the subject is soft, then cropping is just revealing flaws in the image. Hence I want to have good, fast focus on a small object in the frame. And sometimes the object is moving rapidly, making focus more difficult. That is the problem I would like addressed by new models: faster, more accurate spot focusing.

Canon recently announced a 35X Superzoom model, the SX30. I have a friend who tried the Olympus 30X Superzoom and he thought its main problem was focusing accurately when zoomed to its maximum. He showed me some photos and it appeared that about 2/3s of them were too soft to be of any use.

My opinion of the status of the major aspects of DSLR photo taking:

Shutter Speed: no issues
ISO: no issues
Noise: mostly under control
Autofocus: continuing difficulties
Focusing Speed: problem in low light, regardless of the ISO capability
Colour Depth: under control
Dynamic Range: continuing problem (I don't like the concept of adding
frames together. Why can't the processor boost the signal in some areas
and lower it in others within a single frame?)
 

·
Registered
Joined
·
1,006 Posts
The biggest change coming over the next few years is not related to MP or zoom range. We are starting to see the transition to mirrorless cameras. This process will take many years but Photokina announcements show the direction.

- Olympus announced that they will not be introducing any new 4/3 lenses and will focus on m43 going forward.
- Sony announced that the A700 (highend APS-C) replacement will have an EVF, effectively leaving only the recently introduced A560/A580 as the last OVF APS-C cameras. Sony has said that the A560/A580 will not be replaced.
- Panasonic introduced the GH2 (no mirror) with an AF speed of .0099s, which effectively equals or betters most DSLRs. The GH2 can now also support 5fps with continuous AF. We also know that the GH2's Contract Detect AF is more accurate than the Phase Detect AF in most DSLRs.
- In 2011 we will see mirrorless cameras from Canon, Nikon, Panasonic


A year ago it seemed like the concept of a mirrorless camera replacing a DSLR was years away, but now with Panasonic and Sony focusing in this area we are seeing it advance very fast. With the GH2 it seems the AF speed is pretty much there and when combined with the AF accuracy you can see why Sony has been making the switch. The EVF still needs to improve, but the GH2 has made significant improvment there as well (60p refresh, multi-aspect, better colours). And remember that an EVF has many benefits over an OVF like WYSIWYG, larger than comparably priced OVF camera, live histogram, and other information superimposed on the image. I think we are only a year away from the advantages of an EVF overcoming the advantages of an OVF for most applications.

One of the biggest impacts of going mirrorless is the requirement for new mounts. This will have an impact on the future developement of lenses for existing mounts, especially at the low to midrange. When Canon, Nikon, Etc went from Film to digital they kept their mounts for backward compatability, but this is no longer the case with mirrorless. If the low to midrange DSLR bodies get replaced with Mirrorless cameras then the DSLR will end up being supported mainly for pros and the support of legacy lenses. In the past a lens would always be supported with new bodies, but going forward the selection of bodies for legacy lenses may be limited to the high end. I know this will impact my decision when I decide to upgrade from a bridge camera and invest in lenses.

PS. I don't think Sony's Transluscent mirror is a long term answer to the traditional DSLR. It looks like the GH2 can equal or better the Phase Detect AF in the the A33/55 and likely provide better accuracy. The 10fps on the A55 sound neat but the AF cannot keep up in this mode so it is limited. Once Sony has built up their lens selection on the mirrorless e-mount the AF speed will definitely be resolved by then. So what will happen to their A-mount bodies? Sony doesn't have the same following in the pro market like Nikon/Canon so body support may be very limited or dare I say they pull out at some point.

Lots of changes are coming.
 

·
Registered
Joined
·
5,463 Posts
My question is when do bayer based sensors run out of steam.

As a semiconductor guy I think the Foveon (now Sigma) approach has a lot of merit because the light capturing area is inherently much larger and there is no need for micro-lenses AFAIK.
 

·
Registered
Joined
·
1,006 Posts
Other than from a marketing perspective, I don't see the point of advancing Sensors to support more MP. Most lenses cannot resolve the 16mp in the new mid-level DSLRs and the vast majority of people don't realize this. Higher iso is good but once you get to ISO3200 how much more do you need? ISO3200 with a reasonably fast lens will allow you to take indoor shots at 1/60s in most environments without flash.

I hope there is a backlash against more MP that causes the manufacturers to rethink this (ie. like the LX3/S90 sticking to 10mp). 16mp on a APS-C sensor seems like a reasonable area to stop as it still provides the additional resolution needed if people decide to upgrade their lens (ie. we are talking $1k+ lenses).

The future advances will come from improved lens motors (required for fast contrast detection AF), processing power (GH3 has 3 processors), better batteries (EVF in m43 and A55 are power hogs) and software.
 

·
Registered
Joined
·
196 Posts
Sigma's SD1 is technically 46 megapixels, but you'll only see that fully utilized in a very colourful scene. It's interesting because the foveon tech actually doesn't require more resolving power and at worst, even if you take a piture of a plain white piece of paper, it's still 15 megapixels. I'm eager to see what this camera is capable of, especially the dynamic range.

The move away from mirrorless I view as a strategy to compete against Nikon and Canon without competing directly with them...if that makes any sense. The a33/a55 are not the best examples of what can be done, but they are also first generation tech and they're also quite inexpensive for what they are capable of. Panasonic's claims with the GH2 are quite bold, it will be interesting to get a direct comparison with both mid-level and high-end dSLRs. There's a lot of turmoil within the Sony camps too, many were expecting...nay, demanding an a700 replacement to be announced for photokina. All that's been said is that it's still coming and it's an SLT design and there are a lot of frustrated folks moving away as a result.

I like seeing all this competition, the Sigma announcement was really surprising, I love what Fuji has done with the x100, Pentax delivering not one but two very competitive higher end cameras is great and Olympus hasn't forgotten their dSLR contingent with the e5. Individually, these manufacturers aren't really a threat, but together they've got to be holding a lot of influence with what's going on behind the scenes at Canon and Nikon.
 

·
Registered
Joined
·
520 Posts
I'm quite happy with my 16MP full-frame sensor on my 1Ds MkII. Love it. Nice & crisp. Buttery Smooth DOF. Useable ISO6400 would be handy once or twice a year but, for the most part, the only issue (aside from carrying a 10lb brick) is the itsy bitsy Digital Rebel sized LCD.

I agree - no more resolution. I can crop a 4x6 out of 12% of my sensor area at 100% quality at the native resolution of my photo printer. I don't do fine art prints and if I did the new Pentax would be a killer.

Less weight, more sensitive sensor and a higher bit depth would be nice.

What I really want s a G12 sized body that is a rangefinder/EVF with a Leica Screw Mount lens so I can put some of my 50's & 60's glass on a modern sensor.
 

·
Registered
Joined
·
520 Posts
My question is when do bayer based sensors run out of steam.

As a semiconductor guy I think the Foveon (now Sigma) approach has a lot of merit because the light capturing area is inherently much larger and there is no need for micro-lenses AFAIK.
The Bayer approach is dying as I see it. It's essentially a hack. The first manufacturer to come up with a large sensor at a reasonable price will get big market share.

I get comments all the time about how my photos look so nice. Most of the time it comes down to when doing portraiture having a full-frame DSLR with bright lenses I can melt the background.

You just can't do that with a P&S with a sensor that is 1/4"
 

·
Registered
Joined
·
1,986 Posts
As far as higher ISO sensitivity goes, I think there's a lot of room for improvement. Not for 'normal' day to day applications, perhaps, but when shooting indoors with available light. There are a great many dim indoor live music venues where F2.0 @ISO3200 doesn't even come close to getting a decent exposure - an extra 4 stops would be really nice i.e. I'd be quite happy with something around ISO51200 with s/n ratios that are comparable to modern day ISO1600 images.
 

·
Registered
Joined
·
521 Posts
Discussion Starter #9
Interesting Comments

My own biases resulted in me ignoring mirrorless cameras. I have been doing more and more wildlife photography and the major investment (and weight) lies in the lenses used.

Hence whether or not a body has a mirror is largely irrelevant to me. I want fast accurate focusing and strongly prefer looking through a viewfinder, not at an LCD screen. If the camera has an EVF and said viewfinder can handle dark scenes with aplomb I am satisfied. However, if such a camera results in a new type of lens mount, making my lenses non-functional, then I am uninterested in the camera.

Sony seems to be selling their mirrorless cameras by emphasizing their smaller size. Again, if you are carrying a 600mm lens, that size is irrelevant. (Maybe Sony will start selling a fast 600mm lens with the camera built into it. Given the cost of the lens adding a camera to the body will make the lens mount irrelevant. Just make it so that sensors can be replaced as required.)

I too was surprised by the new Foveon model. I thought that they had given up as their resolution was just not good enough. I wonder if because of not employing a filter, more light gets through to the sensor. If so, then if the camera used the same techniques as normal DSLRs, it should be able to function in very low light. One of the technical problems that pushed the manufacturers to CMOS chips was speed. Namely how quickly could a pixel clear the information it had gathered in order to take a new photo. Memory latency it is called in PCs. I wonder about the functional speed of Foveon chips. And is latency a function of light frequency, resulting in different pixels requiring a different period of time to clear the signal? I.e., does a "green" pixel process as quickly as a "red" pixel in a Foveon? And recall that light at dusk and light in shade has radically colour balance than light in normal sunlight. Couldn't a Foveon automatically adjust the gain for the different frequencies to create perfectly balanced photos?

I like the idea of a 12-stop dynamic range, such as boasted about by the new 80 MP Leaf sensor. When I am walking around I have had to adjust my shutter speed from, for example, 1/4000 second to 1/160 second after covering about 20 meters of distance. Such is the impact of moving from a meadow to a covered forest. When such lighting is dappled, then a sensor with a 12-stop range would be useful. Perhaps combining frames to gain dynamic range works for static images, but not with quickly moving subjects. So I prefer HDR to be embedded in the sensor, rather than in the processing of multiple frames.
 

·
Registered
Joined
·
5,463 Posts
I too was surprised by the new Foveon model. I thought that they had given up as their resolution was just not good enough. I wonder if because of not employing a filter, more light gets through to the sensor.
The initial problem Foveon had was in the pixel wars. They didn't do a very good job of countering the pixel counters.

Because the capture sites are stacked vertically there is more area to capture light so everything else being equal they should be better at low light. Of course the "permeability" of the silicon for the different colours comes into play as well as the "gain", for lack of a better term of each capture layer. Those those are all easily addressed in the A/D and post-processing steps though.

As far as speed goes, that's a function of how fast they can drain the charge in each photosite. In this scenario the large sites could limit the shooting speed since they would take longer to drain unless you addressed that with a larger current path.
 

·
Registered
Joined
·
196 Posts
I just read somewhere today that the SD1 also doesn't have an anti-aliasing filter, if that's true I'm going to find myself reconsidering just how limiting the Sigma mount really is.
 

·
Registered
Joined
·
1,894 Posts
There is one non-obvious use for excess mega-pixels, namely binning. I'm a total novice at the local camera club, but we all have to start at the bottom. One restriction for competition submissions is that they must not exceed X x Y = 1024 x 768. Either or both dimensions can be less, but not more. So why do I shoot full-size 4308 x 2868?

I bin the image by a factor of 4:1 using imagemagick, resulting in a 1077 x 717 image. The X (width) needs a bit of cropping, but binning by 4:1 makes an ISO 6400 image look like ISO 1600. This is REAL noise reduction BEFORE USING DE-NOISING SOFTWARE. See http://www.noao.edu/outreach/aop/glossary/binning.html for a beginner's explanation of binning. It's oriented to astronomy, night photos. But the principle is valid in any low-light situation.

There's no such thing as a free lunch. I'm trading off pixels for a cleaner image. Since I'm not allowed to submit a larger image than 1024 x 768, I may as well get some use out of the extra pixels.
 
1 - 12 of 12 Posts
Top