July 22, 2014, 04:33:07 PM

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - jrista

Pages: 1 ... 75 76 [77] 78 79 ... 248
1141
Animal Kingdom / Re: Show your Bird Portraits
« on: January 16, 2014, 07:38:24 PM »
First day of trickery.  They love my stump! ;)

I chose 300 X1.4 and 1250th and ISO 800  F5.6 typical

Jack

Very cool.  Think I'll have to try something like this.  At least my wife wouldn't have to worry about me being away from home all weekend.

Don't forget, you can create setups out in nature as well. I get a small selection of Colorado's 400+ bird species in my yard. I have to go elsewhere to photograph a greater species diversity. Alan Murphy has a bunch of tricks for setups in nature as well. I keep meaning to try a setup to get some good photographs of the local Pheasants (which are notoriously difficult to photograph as they are always hiding in the grass and brambles.)

1142
EOS Bodies / Re: Will Canon ditch the AA Filter?
« on: January 16, 2014, 07:29:44 PM »
I know for a fact that LR uses a form of AHDD. DPP tends to produce sharper results, however it is more susceptible to aliasing. LR tends to produce really clean edges, but it also produces slightly softer results.


Hmm I've actually found that I can get more detail out of ACR than DPP (although I haven't used DPP since the convolution stuff, but barely at all) and ACR seems prone to created stair-stepped fine lines from some lenses.

ACR seems to be much more prone to color moire aliasing than DPP, but DPP seems to be much more prone to zipper artifacts. ACR is also prone to severe stair-stepping jaggies at times and it can be really bad at timed with say orange leaves against a blue sky. C1 avoid the stair-steps but seems prone to dotted line type artifacts (where a solid, fine line may turn into a dotted line or a line with spurious stray pixels hanging off or certain solid patterns turn into randomly dotted checkerboards). Each definitely has some issues.

I'm not saying that ACR can't get more detail...just that it's not quite as intrinsically sharp as DPP (without any adjustments beyond simply demosaicing.) The aliasing issue with DPP is pretty consistent, doesn't require any specific kind of scene...sharp edges are usually aliased a bit, regardless. In my experience, fine edges and lines with ACR are usually very clean and crisp, without the endemic jaggedness that occurs frequently in DPP. Sure, there are some difficult cases that fall outside the realm of normal for any algorithm, but AHDD is one of the better demosaicing algorithms, and DPP clearly does not seem to use it.

1143
EOS Bodies / Re: Will Canon ditch the AA Filter?
« on: January 16, 2014, 07:25:36 PM »
Moire exists outside of the world of digital cameras. Think of looking through two sets of railing, bird feathers overlapping and so on. If you're there looking at it with your own eyes, the moire is visible. Therefore this particular type of moire is not camera induced.

It very well can be. It's not intrinsically there, if the eye sees it, it would only see it at certain particular scales of distance.

Moire is the result of pattern interference. It doesn't really matter the scale, the effect is scaleable. If you have two crosshatch lattices in your yard, and you were able to line them up relative to each other, the two would create a moire pattern. All that is necessary for moire is for patterns with similar frequencies to interact and interfere with each other. Palm fronds crossing over each other, two nets overlayed on top of each other, bird feathers crossing over each other...all of these create moire. Moire is a consequences of waves in nature (no, not water waves), waveforms interacting and interfering to produce an entirely different "output" waveform that does not resemble any of the originals.

Moire doesn't come from light wave super positioning and interference. That doesn't even make sense, since you can print out large thick lines on a piece of paper and create it, that is pattern frequency and scale interaction and it's all positive additive.

I never said moire came from light waves at all. It comes from WAVES. Waves exist in all kinds of spatial media, it implicitly has nothing specifically to do with just light.

And I mean you can photograph some suit with one sensor from one distance and if the scale is right you get nasty interference but then move the camera back or forward and it changes or goes away so it's not like you can say the suit had some intrinsic moire. It was the combination of the pattern from the suit projected onto the sensor at specific distance that did it. So in that sense it is not intrinsic or scalable. I mean you have created the moire in the captured image maybe it's sort of scalable at that point within reason.

Moire can be created by patterns in the image interfering with the pattern of the sensor, obviously. That's why we have AA filters. Certainly, you can move closer or farther from your subject, and change the moire effect created, assuming you have that option given the composition your aiming for. That is just one means of creating moire, though. Moire does not need a sensor to occur...any repeating waveforms with the right frequencies that can interfere with each other will produce moire. You can plainly see it with your own eyes, without the existence of a camera, in countless real-world situations. If you photograph such a thing, the moire will be in your photo, in that case not as a consequence of interference with the sensor, but simply because it existed in the real world.

In the case of my bird photos, there are a few spots of moire, because it existed in reality, not as interference with the sensor array.

1144
EOS Bodies / Re: Canon 7D - Vertical Banding Noise
« on: January 16, 2014, 04:02:33 PM »
Not good. Looks like one of the readout channels on the sensor has gone bad.

At least one. The 7D has eight channels...looks like maybe two of them went out, given the spacing of the missing columns.

1145
EOS Bodies / Re: 7D Mark II on Cameraegg
« on: January 16, 2014, 03:58:23 PM »
I am hoping the 7D2 is to the 7D, as the 5D3 was to the 5D2 and is compelling enough to make me want to upgrade. I am specifically looking and hoping for a sizable improvements in DN, High ISO Performance and the 5D3 AF system or similar.
I am so disappointed with the 7d by today's standards that any improvement will be worth the upgrade.  Just show me where to sign...

The problem Canon has with the 7DII is that it will certainly rob sales from the 1Dx. If 10 fps @ 18mp with 61 point AF is true then it really will be a 1.6x cropped 1Dx. I suspect that Canon will leave the IQ much the same as the 70D, it's got to hold back on something for this camera.
Naturally we all want a camera with less video compromises (the 7D has a really strong AA filter), far better noise handling threashold and a DR to match the current Sony Exmore CMOS sensors. But I seriously doubt that Canon will invest in sensor tech for this camera and I think they will brin in the new tech for the 1Dx replacement or it's high MP cousin.
I don't think  the 7D II rob sales from the DX. I've been saving for the 1 DX for many months now and on the verge of the getting one. I did have a fleeting thought maybe I wait for the 7D II I can then purchase that along with the 5D Mark III. I'm thinking maybe I can have the best of the both worlds but I held that 1DX again and felt the solid build. There's no doubt I'm getting the DX. The 1 DX and 7DII will cater to two different markets IMHO...

I occasionally wonder if I will be hit with the 1dx bug.  My logic in the past has been... if you have the money and your lenses are really good, the logical step is to improve your body. 

But there are also times when I use my daughter's old xti and I think... this generations old body can get me wow photos too... so why do I have a $3000 body and by extension, why would I ever need an $8000 body.

And I know the answer... because I can't use the xti all the time with a tripod in ideal light.  But it does still bother me just a little. 

If you don't mind me asking, what does the 1dx bring to the table that a 5d mkiii can't provide?  I know the specs, but I'm just curious what you value to be worth the extra cash?
Sure, I want to be able to expand into sports & wild life photography. Don't get me wrong the 5D Mark III is a great camera and it will handle most of my needs but it just doesn't have the blazing fps. I pondered this for countless months and I won't be purchasing aonther body for a while after this one. I currently own a 5D Mark II.

The 7Ds AF system has an intrinsic jitter that negates a lot of it's FPS advantage. Even when locked onto a subject, there are small changes in the actual AF position every single frame, which can kill that "one best frame". Overall, the keeper rate with the current 7D falls around 4-5 per second, at best. The 5D III has a vastly superior AF system, putting it in the same bucket, and possibly even beating the 7D with a consistent 5-6 keepers per second (it's AF system rarely misses and doesn't seem to jitter.)

Unless Canon actually puts a better AF system in the 7D, and guarantees that whatever AF system they do put in it doesn't have that jitter, the extra FPS isn't going to be all that much of a lead over the 5D III.

1146
Animal Kingdom / Re: Show your Bird Portraits
« on: January 16, 2014, 01:14:11 AM »
Jack, Thanks!  If I had stopped shooting to take the 1.4X off, I might have gotten everything in the frame.  Was really concerned that the required movements would cause the hawk to fly off but after it was over, I believe it might have been possible.

I say stick with your instincts in the field. Hindsight doesn't tell you as much as having that in-person presence does. I've often been in the same situation, thinking I could have taken off the TC...but usually, the birds I photograph just don't sit still for long, and taking the time to switch the TC in the middle of photographing an interesting sequence usually results in losing the moment.

1147
EOS Bodies / Re: Hybrid Viewfinder Coming To Canon DSLRs? [CR1]
« on: January 16, 2014, 01:12:35 AM »

It's a Bower. I think it's the same generic brand as Rokinon/Samyang.

It may not be the Bower TC that causes the IQ loss. I mean, it will cause some, but the 100-400 sucks even with the EF 1.4x III (which it doesn't even function properly with), and while it functions properly with the much-loved Kenko Teleplus Pro 300 GDX 1.4x TC, the IQ still sucks (a little bit more than the Canon 1.4x, but the differences aren't huge.) I think it's just that the old 100-400mm lens design was built in the film era, at the early dawn of the digital era, and the bar for quality wasn't as high back then. It is most definitely a softish camera at f/5.6 and f/6.3, and only really starts to sharpen up by f/7.1 and f/8. With a TC, you would be at f/11, which imposes a significant hit on either shutter speed (which increases softness from camera shake) or ISO (which packs on the noise, especially on a 7D).

It is possible that your $80 Bower TC is just fine, and that it just doesn't pair well with 100-400 (because, well, NO TC pairs well with that lens. :D )
There's probably something to do with the age of the 100-400 and its dislike of TCs, but I've tried the Bower TC on my old 28-135, and the TC definitely degraded the sharpness there too.  Maybe I got a bad copy.  I imagine the Kenko brand is a tad better than Bower.  I hear the Bower/Rokinon/Samyang conglomerate makes fairly good lenses, so maybe they just don't make good teleconverters.  It's probably very difficult to optimize a teleconverter for a bunch of very different lenses.

The Kenko is decent, but it is definitely not as good as the Canon TCs. It allows just barely visible improvements when attached vs. when not attached. Subjects are definitely larger in the frame, but you don't get the same kind of increase in overall detail as with a Canon TC.

Something else I've noticed with the Kenko TC...boke circles look TERRIBLE. They have this funky warped star effect which just looks rather bad, so I don't really use it much anymore. It's great though, for people who want f/8 AF on camera bodies, like the 7D, that don't normally support it (so the boke issue just doesn't matter in those cases). It does allow f/8 AF, and in good light, even the 100-400 will focus automatically, albeit slowly.

If you have a good lens, and a body that supports AF at the smaller apertures, get a Canon TC. No question they offer better quality. If you have a body that does not support f/8 AF and you need it (probably best with the 400/5.6 L prime), the Kenko is the best bet. (Actually, the Kenko MC4 seems to produce better IQ overall than the Kenko Pro 300 DGX, so I actually recommend getting that one...I simply couldn't find one for sale when I bought mine.)

1148
EOS Bodies / Re: Hybrid Viewfinder Coming To Canon DSLRs? [CR1]
« on: January 16, 2014, 12:22:11 AM »
Daniel, I'm no expert to believe! ;)  One thing for sure though is that the 1.4X and 2X if they help you to fill your frame, are better than cropping.  I've proven that with my 300 2.8 II and converters III.

I believe what some like me are hoping for is more reach while still being able to keep the portability of the 300 plus converters, and for AF improvement.  While I might be able to scrimp and afford a 600, I would be seriously challenged to take it where I would most want it.  Thus, I want to try out the elusive 7DII.

If you're thinking wide, go full frame for sure.

Jack
A lot of argument over the relative value of teleconverters and crop/ff depends a lot on the lens.

If you use a teleconverter on a soft lens, you can decrease the resolving power. If you use it on a sharp lens you will increase the resolving power.

Likewise, a soft lens on a FF camera may out-resolve (or at least be close to) that same soft lens on a crop camera, yet with a sharp lens it may be the other way around...

In other words, if you are going long, the lens quality is more important than FF or Crop.

And Jack is definitely right about going wide.... FF all the way! The wider the angle of the lens, the harder it is to design without distortion and the harder it is to make it sharp. If you wanted the same angle of view on a crop camera as 24mm on a FF camera, you would need a 15mm lens. With the same level of design and materials you can not make a 15mm lens as well as a 24mm lens, so just on the glass, FF wins in IQ.....plus there is all the usual stuff about less noise, higher ISO, etc etc...


This is very true: certain lenses are known to disagree with teleconverters - the 100-400L is a good example of that, for multiple reasons.  The teleconverter quality also matters - a cheap teleconverter can be worse than cropping a photo.  I have an $80ish TC that actually degrades the quality of my 100-400 - a total waste of money.

I have a Canon 1.4 mkii teleconvertor.  Just as a cautionary tale... mind telling people which brand the $80 TC is?
It's a Bower. I think it's the same generic brand as Rokinon/Samyang.

It may not be the Bower TC that causes the IQ loss. I mean, it will cause some, but the 100-400 sucks even with the EF 1.4x III (which it doesn't even function properly with), and while it functions properly with the much-loved Kenko Teleplus Pro 300 GDX 1.4x TC, the IQ still sucks (a little bit more than the Canon 1.4x, but the differences aren't huge.) I think it's just that the old 100-400mm lens design was built in the film era, at the early dawn of the digital era, and the bar for quality wasn't as high back then. It is most definitely a softish camera at f/5.6 and f/6.3, and only really starts to sharpen up by f/7.1 and f/8. With a TC, you would be at f/11, which imposes a significant hit on either shutter speed (which increases softness from camera shake) or ISO (which packs on the noise, especially on a 7D).

It is possible that your $80 Bower TC is just fine, and that it just doesn't pair well with 100-400 (because, well, NO TC pairs well with that lens. :D)

1149
EOS Bodies / Re: Canon 7D - Vertical Banding Noise
« on: January 16, 2014, 12:13:05 AM »
If they are going to charge $1,000+ to change the sensor, plus my cost to ship it there, I won't bother sending it in to Canon.  So, I posted that top photo (red car) online and a link to it - and sending that link to Canon Canada tech support to see if they will look at it and give me at least and idea of what they think the problem is.  No sense sending it in if it's $1,000+.... I'll just buy another 7D.

I've been considering selling my 7D to build up some cash for a 5D III. If your interested, let me know. I'll let it go for a decent price under $1000...it has some cosmetic scratches, but otherwise, the thing is a damnable tank! (I hope the 5D III holds up as well as my 7D has...love the durability of this camera!! :D)

1150
EOS Bodies / Re: Canon 7D - Vertical Banding Noise
« on: January 16, 2014, 12:00:12 AM »
An old PowerShot of mine had a very similar banding issue - I ended up sending it to Canon.  It needed the entire sensor and lens assembly replaced.  Canon's repair costs are actually surprisingly reasonable for what they do to the gear, I'd actually recommend just sending the camera in.

The cost wouldn't be reasonable if it was more expensive than a brand new camera....or for that matter, the majority of the cost of the camera (the 7D sales for $1200 very frequently....so even the low end $1000 cost simply isn't worth it, IMO.)

1151
Animal Kingdom / Re: Show your Bird Portraits
« on: January 15, 2014, 11:59:05 PM »
Some great ideas for the yard, or anywhere for that matter, Jack and jrista.  I actually laughed reading your post Don(roughing it).  I've gotten most of mine by backing into a bush but a full blind would probably be better.  Some of you might have seen this video by Cornell.

...

Nice Mandarin closeup Harv!  Supposedly we have one or two here but they were imports :(

Loving the snowies, ftico and IslanderMV.  Still a tough call, 80F clear weather or snow, freezing temps and snowies.

Awesome video! Hadn't seen that before, guy did a great job. I know some Spotted Sandpipers nested and hatched chicks around the Cottonwood Creek Wetland near me last year. I should set up my blind out there this year, and see if I can capture some good shots. I never brought my blind last year, and man, daddy piper did his job well. I figure if I could get set up before sunrise, I could probably get some good shots and maybe video of the pipers around Cottonwood Creek this year.

1152
Animal Kingdom / Re: Show your Bird Portraits
« on: January 15, 2014, 11:51:57 PM »
jrista, now you got that off your chest, you'll feel better. ;)

I looked for that chair and found one at Canadian Tire that looks the same, around $100.  What's your opinion of that particular chair and is there anything else that might be preferable?  That's better than the open lawn chair that I parked at "my" pond during the summer (actually my acreage backs onto county reserve so technically it's not "my" pond (another confession). ;D

Jack
I picked up one at Canadian Tire that is about 4 feet square on the base and sits about 5 feet high.... It is sitting out in the yard with a comfortable folding chair in it... and I must also confess that there is a long extension cord and a small space heater.... takes the edge off of things when you are shooting below -10.... and it is also within WiFi range of the house... (Roughing it is for amateurs)... I was sitting in it on the weekend trying to catch chickadees in flight (DARN THAT'S HARD!!!) and this lovely lady ran through the yard..

Haha! Indeed! :P I do pretty much the same thing...I have a little space heater in mine on the cold days, too. Actually, if you put small stones on the bottom edge of the tent, and zip up tight around the lens, it actually holds heat in pretty darn well, even when the temperature is -10.

As for the Chickadees in flight, you should get Alan Murphy's eBooks. He has some good setup tricks in there that helps a LOT. All the best songbird/passerine BIF shots are setups...it's pretty much the only way to do it. The trick is getting them to land a short way away from "the treat"...seed, berries, suet, whatever...so that they will make that short little "flight" between the initial perch and their goal in a known spot. Then, you basically just point your lens towards one spot, leave it there, and fire away the moment the bird looks like it's going to take off from that initial perch.

Of course, that doesn't necessarily mean you'll get any shots. I have yet to create a setup where birds will land first on my special perch, then fly to the treat bar... They always seem to find a more amenable perch from which they launch their attack on the food, so I don't have any passerine flight shots I consider keepers yet. I think part of my problem is space...I only have a quarter acre of land, and everything is pretty tight and close. I want to give some setups out at the nearby state parks a try, see of I can lure some ground fowl and some passerines into a handy little setup and get some better shots.

1153
Animal Kingdom / Re: Show your Bird Portraits
« on: January 15, 2014, 10:01:36 PM »
jrista, now you got that off your chest, you'll feel better. ;)

I looked for that chair and found one at Canadian Tire that looks the same, around $100.  What's your opinion of that particular chair and is there anything else that might be preferable?  That's better than the open lawn chair that I parked at "my" pond during the summer (actually my acreage backs onto county reserve so technically it's not "my" pond (another confession). ;D

Jack

Your lucky to have land. I live in the suburbs, so I don't have nearly as much space for a really good bird setup, let alone a pond! :P

As for the chair, it's excellent. There are little huts that fit multiple people, tall huts for standing in, etc. But this chair is really the best option. It can be a bit of a pain to get the telescope with a big white lens to fit into initially, but once it's done and you have the lens pointing out the front window, you can sit in it for hours and the birds really don't know your there. You can even make sounds, talk on the phone, whatever...so long as your inside that "odd looking big bush", they just don't care. The moment you poke your head out, though, they all scatter.

I spent $80 on mine on a sale at NatureScapes.net. I think the regular price was $99. Definitely worth it, though.

1154
EOS Bodies / Re: Will Canon ditch the AA Filter?
« on: January 15, 2014, 09:25:04 PM »
I realize this is the CANONrumors forum, but we seem to have some people engaged here who understand AA at a fundamental level and I'd like to pose something to them.

Fujifilm's X-Trans sensor (APS-C sized, 16mp) doesn't use an AA filter.  They claim by using a different Bayer filter layout, they don't need an AA filter.

http://www.fujifilmusa.com/products/digital_cameras/x/fujifilm_x_e2/features/index.html

Blowing smoke or do they have a point?


I think it is definitely a legitimate approach, assuming it's real (more on this caveat in a minute). Aliasing (of which moire is a type) occurs when two regular repeating patterns interfere with each other. For example, the even grid of fabric, when resolved by a lens such that the threads of the fabric resolve to about the same size as a pixel, is what opens up the opportunity for aliasing. The regular pattern of the threads in the fabric, even though they may resolve to similar dimensions as the sensor's bayer array, will never line up perfectly. Since a CMOS image sensor resolves discrete points, rather than a continuous analog signal, the sensor ends up "aliasing" real-world information. This image from wikipedia demonstrates (they used to have an animated one, seems to be gone now):



Assume the top third is the sensor array, the middle third is the fabric your photographing, and the final third is the result of the two patterns interfering. The result is effectively nonsense. It doesn't resemble reality at all, most certainly doesn't resemble the fabric you were photographing. The problem with modern cameras is even worse...you have an uneven distribution of color pixels, which results in not just an interference pattern, but a color interference pattern (and they can be quite hideous, and take on a whole variety of forms, depending on what kinds of patterns are interacting, so there is no well-known algorithm that can easily clean moire up once it's been baked into your photos...not without severe degredation in image quality.)

Fuji's X-Trans claims an interesting in that, instead of using the most sparse approach to interpolating the raw sensor data, they use a very greedy approach to interpolating the raw sensor data. Moire occurs when analog signal patterns as resolved by the lens interfere with the bayer array. Normally, each bayer pixel is roughly the same area as an output (screen) pixel, input pixels and output pixels are at roughly the same frequency, so that interference shows up when you view a RAW image on your screen in something like Lightroom. X-Trans interpolates a 6x6 matrix of pixels, meaning that one output (screen) pixel is blending information at a larger frequency...moire is INTRINSICALLY blurred away. By interpolating a 6x6 matrix of sensor pixels, X-Trans is technically reducing the maximum potential resolution...by a factor of three...but technically speaking the results would be superior.

Now, there is a caveat I mentioned before. X-Trans could be better, IF that is actually what's happening! I researched the X-Trans a while back, and could never reconcile some of the specifications. Fuji claims the sensor has 16.3mp effective, and 16.7mp actual pixels. The full size RAW output dimensions are 4896x3264. Well, first, if that is the dimensions of a RAW image, that is actually 15.9 megapixels, not 16.3 megapixels. Second, with 15.9, 16.3, or 16.7 megapixels, there simply aren't enough pixels to perform 6x6 interpolation. This has always thrown me for a loop.

Fuji's images are a bit misleading about how bayer interpolation works as well. They use the following diagrams:



On the left is what Fuji claims is standard bayer interpolation. With modern RAW editors, AHDD, or Adaptive Homogeneity-Directed Demosaicing...or some close variant thereof, is pretty standard. AHDD uses a much more complex approach to bayer demosaicing that aims to maximize the information recorded in a standard bayer sensor, minimize stairstepping, minimize color artifacts, and minimize aliasing. AHDD aside, however, true "standard" bayer demosaicing is actually like this (this is an animated GIF image I created some time ago to describe bayer demosaicing):



Watch the image above for a little while, as it counts all the groupings of 2x2 sensor pixel matrices that are actually involved in the most basic form of bayer demosaicing. Notice the multiple overlapping. It isn't each discrete 2x2 matrix of pixels that are demosaiced without any overlap (as in Fuji's image)...it is every possible INTERSECTION of all neighboring pixels that is demosaiced, WITH overlap. A lot of information is reused, so every last scrap of high resolution detail is preserved. Luminance detail is very high with standard bayer, while only color information is spread around more and "blurred" a bit.

Fuji doesn't really explain in all that much detail exactly how they interpolate their X-Trans bayer array. Do they overlap a bit? Given how they rearranged the layout of red, blue, and green pixels, there don't seem to be as many options for overlapping. Even assuming they do overlap...with a 6x6 matrix interpolated for each output pixel, overlapping would greatly soften the output image, both in luminance and color.

The only way I can reconcile how Fuji explains their interpolation process with the megapixel counts and RAW image dimensions is to assume that when they say "megapixels", they mean output pixels, not sensor pixels. That would mean the sensor actually has pixels that are 1/6th the pitch, and that the sensor is 575 megapixels of super tiny pixels, which would be high in noise and low in color fidelity. The high noise/low color fidelity wouldn't matter like it does in a standard bayer, because they are interpolating 6x6 of them for each output pixel. Problem is...575 megapixels? If Fuji had actually achieved that...they would have said something...

So, I'm left with this conundrum. I can see the benefit of interpolating discrete non-overlapping 6x6 pixel matrices. It would certainly eliminate the need for an AA filter. It brings in more color information per output pixel, so color fidelity is higher. It brings in more light per output pixel, so noise would be lower. Theoretically, it sounds awesome. However, with only 4896x3264 pixels in the sensor...the interpolated output resolution would have to be 816x544. That also doesn't make any sense.

I suspect Fuji has some more complex approach to interpolating their RAW sensor data that isn't nearly as simple as a 6x6 matrix of "more random" red, blue, and green pixels. What that interpolation algorithm is, I cannot say. Given the way they arrange pixels, I think it is entirely possible they are interpolating 3x3 matrices of pixels in some AHDD-style demosaicing process that more richly factors in luminance information from all pixels, and cleverly interpolates color information so as not to blur too much. I do think there is some blurring...in a lot of X-Trans sample images, I've often seen halos around things, or softness where one would expect crispness.

X-Trans has really great theory. In application, I don't think it quite lives up to it's claims of critically sharp detail (not on the level of something like the 5D III, 1D X, or D800 for example), however it does seem to have very low noise at high ISO, and I can't say I've seen any moire or aliasing in any of the sample images I've viewed in the past.


Jrista, your demosaicing gif got me thinking.

Canon must be using a slightly different method. Here is a 100% crop of a noisy pixel off my 5D2. When the image is initially being processed (before final resolution is shown) the dot is originally green, then turns grey as seen here.
It looks to me like the green pixels are being average out separately from the rest, it would be interesting to see if lightroom and other editors produce similar effects from hot pixels.





My GIF just demonstrates the most basic form of bayer demosaicing. It is highly likely that DPP is using something more advanced (although what exactly, I don't know). I know for a fact that LR uses a form of AHDD. DPP tends to produce sharper results, however it is more susceptible to aliasing. LR tends to produce really clean edges, but it also produces slightly softer results.

There are lots of trade offs to be made for using different forms of demosaicing. There is also "SuperPixel demosaicing", which uses groups of 2x2 RGGB pixels in NON-overlapping space. You get less resolution and can still suffer from mosaicing, but higher color fidelity and better sensitivity per output pixel.

1155
EOS Bodies / Re: Hybrid Viewfinder Coming To Canon DSLRs? [CR1]
« on: January 15, 2014, 07:21:59 PM »
Although I haven't longed for a hybrid viewfinder, I can see its benefits as a stills shooter as well. Particularly if they offer a "Heads Up Display" sort of arrangement, much as Fuji has done in some of the X series. For example, a  transparent, live Historgram overlaid over a corner of the optical view might be useful in challenging lighting conditions.  Only time will tell if this is the sort of thing they plan to implement into this new hybrid viewfinder. 

TOTALLY AGREE!! I've said that very thing a few times in the past...with Canon's existing translucent LCD overlay for their OVFs, they could put in a monochrome histogram, which would be IMMENSELY useful for those times where you can't really take your eye away from the viewfinder, and the basic exposure meter isn't sufficient to gauge proper exposure. I think there are so many things Canon could do with their current OVF technology...would be very cool to see them take it farther.

As for Canon's lack of innovation in sensors, well that may be coming to an end.  Certainly DPAF has shown that they can be inventive and innovative.  Insofar as the technology is more useful for video, some might think that it's a sign that Canon is favouring video over stills, but I don't think that's the case.  We've already been told that using Dual Pixel technology for AF is just the beginning of what it can do, so it could be an interesting 3 or 4 years for Canon if that promise can deliver new tricks for stills. 

I also think it is wise to keep in mind how amazing the original 7D was when it first came out.  Given how the 5D series has moved from strength to strength, I don't think Canon will compromise on the quality of the 7D II, simply because they will lose a golden opportunity to build on an already incredibly popular product line.  If they "blow it" with this product, it will hit them hard where no corporation wants to get hit--their reputation.  Although the bread and butter Rebels may be what pays the bills and keeps the lights on, it's the high-profile "flagship" products that allow them to sell those rebels.  Delores from Idaho may not know which end of a 1DX is which, but she knows there are lots of Canon cameras at the Olympics and as well as in the hands of pro photographers, and that is what makes her decision to buy a T3i, or SL1 (or a little Canon pocket camera, for that matter).   Companies can afford to "blow it" with a low end camera, then replace it with a nearly identical model some months later.  Witness the "oops--T4i had a problem with the grips turning white.  Better add a fully rotating dial and call it a T5i six months later!" situation.  But the same isn't true for a high-end camera.  A bad rap in a premium camera is harder to shake. 

Another point to consider is the time factor.  Given how long people have been waiting for it, Canon doesn't want the 7D II to underwhelm.  It may actually be a positive that it's been so long, insofar as it might mean Canon is waiting till the product is "right" before they release it.  Consider the 200-400 F4 IS with 1.4TC--the wait was so long it became a running gag that it was more akin to Bigfoot than a real lens.  Yet it delivered all that was promised of it when it finally shipped.  I know, for most of us that's a moot point--we can't afford it.  But we all still wanted it to be phenomenal anyway--and it is!  I think it will be like that for the 7D II--except this camera WILL be affordable (if still expensive compared to the other APS-C cameras). 

I've said it before but I'll say it again, I think Canon will make the 7D II (and I hope they choose that name for continuity) will be great, because they don't dare make it anything less than phenomenal.  I could be wrong--in which case it will likely flop and flop badly.  But I still hold out hope that it will be everything the 7D was in its time---and more!  ;D

+1 Great insights. Particularly the points about the great overall leaps forward the 5D III and 1D X were relative to their predecessors. I guess if you apply that logic to the 7D II, which is still a pro-grade DSLR, then that would indicate the 7D II should see a similar overall leap forward. I still fear THE TRICKLE...but maybe that's just Canon's MO for lower end products (which they release far more of far more frequently than pro-grade anything.) Maybe the 7D II will still be what we all hope it will be, plus the Hybrid VF.

(Personally, I am not concerned with timeframe, as you say...the more time Canon takes on the 7D II, the better it should be when it finally arrives. I can get the 5D III in the interim, which I suspect would still be my primary even if I picked up a 7D II in the future.)

Pages: 1 ... 75 76 [77] 78 79 ... 248