rscheffler wrote:
Just came back from a nice fall hike, shot the 90D with the 70-200/4L IS v1. Sure beats carrying around a 1DXII and got great reach thanks to the APS-C crop. A quick review of the images revealed that a 32MP APS-C sensor is very demanding of lenses... I very much like the 70-200 on FF though it's not perfect with respect to across-frame sharpness. The 90D's sensor seems to really amplify all lens characteristics, good and bad. Sharpness is certainly there, but I'm also seeing a lot more softening in corners and/or uneven across-frame sharpness. It seems for some of my lenses, it may mean stopping down into diffraction territory for best overall sharpness, giving up some micro contrast....Show more →
But .. if you're printing to the same size (inches, not pixels), the uber layer of scrutiny isn't going to reveal those characteristics the same as when you're peeping. Kinda like looking at a woman with porcelain smooth skin, then getting out a microscope to inspect the pores in here skin. The micro-viewing will show you the terrain looks different once you put your nose up to it. Your call if you want to look at her through the microscope or from a view that beholds her beauty.
Diffraction territory is kinda like applying makeup to smooth out those pores, when it really may not be needed ... you know, since others aren't viewing her skin with a microscope when she walks out the door.
I read that pretty small article as more a discussion on higher resolution sensors in general, with a nod to the 90D, and less as a glowing testimony to the 90D. Sony has their high resolution sensor already on the a7r4, we just need Canon to start to show their cards a bit in various areas, instead of being so quiet, which makes people a bit uneasy.
Summary of the article in two sentences:
"But the arrival of two newer, higher pixel-count sensors, both of which outperform those 24MP chips in meaningful ways, is likely to herald the end of the 24MP era."
"With the 24MP sensors looking outdated in terms of both stills and video performance, it's likely we'll see more cameras moving to these newer, higher resolution sensors"
rscheffler wrote:
Just came back from a nice fall hike, shot the 90D with the 70-200/4L IS v1. Sure beats carrying around a 1DXII and got great reach thanks to the APS-C crop. A quick review of the images revealed that a 32MP APS-C sensor is very demanding of lenses... I very much like the 70-200 on FF though it's not perfect with respect to across-frame sharpness. The 90D's sensor seems to really amplify all lens characteristics, good and bad. Sharpness is certainly there, but I'm also seeing a lot more softening in corners and/or uneven across-frame sharpness. It seems for some of my lenses, it may mean stopping down into diffraction territory for best overall sharpness, giving up some micro contrast. ...Show more → RustyBug wrote:
But .. if you're printing to the same size (inches, not pixels), the uber layer of scrutiny isn't going to reveal those characteristics the same as when you're peeping. Kinda like looking at a woman with porcelain smooth skin, then getting out a microscope to inspect the pores in here skin. The micro-viewing will show you the terrain looks different once you put your nose up to it. Your call if you want to look at her through the microscope or from a view that beholds her beauty.
Diffraction territory is kinda like applying makeup to smooth out those pores, when it really may not be needed ... you know, since others aren't viewing her skin with a microscope when she walks out the door. ...Show more →
I get what you're saying and one aspect of 32MP that appeals to me is the amount of detail that carries over when downscaling vs. images shot natively at that resolution. That said, when pixel peeping, it's a bit disconcerting that a decent FF lens is showing problems within the area of APS-C coverage. The 85/1.4L IS has generally been a lot better and the 35 IS is also good, though not necessarily great.
On another note, what apps are others using to process 90D CR3 files? I'm still holding out with LR6 perpetual license and have to convert the CR3s to DNGs. Imported into LR and the files look really lifeless compared to the SOOC Jpegs. Of course it's possible to tweak them back to looking good, though I'm not yet really happy with skin tone rendering. I used to use DPP ~10 years ago and loved its color and sharpness, but don't have a desire to return to it now...
On another note, what apps are others using to process 90D CR3 files? I'm still holding out with LR6 perpetual license and have to convert the CR3s to DNGs. Imported into LR and the files look really lifeless compared to the SOOC Jpegs. Of course it's possible to tweak them back to looking good, though I'm not yet really happy with skin tone rendering. I used to use DPP ~10 years ago and loved its color and sharpness, but don't have a desire to return to it now...
I recommend you try DPP again with the 90D. First would be for noise reduction. The noise reduction is much superior to LR. At ISO 6400 you can obtain results that are close to ISO100. At 12800, you will need extra sharpening and will lose minimal details. The biggest issue is likely to be color shifts which can be difficult to correct, depending on the specific image. Next DPP includes Canon's diffraction correction software. I have been very impressed with results. There is very minimal loss of sharpness even at f/20. DPP is slow but I find it worth the effort.
Between DPP and good post processing skills, one might get up to two stops noise reduction with very minimal detail loss. Anything more produces ever increasingly inferior results.
rscheffler wrote:
I get what you're saying and one aspect of 32MP that appeals to me is the amount of detail that carries over when downscaling vs. images shot natively at that resolution. That said, when pixel peeping, it's a bit disconcerting that a decent FF lens is showing problems within the area of APS-C coverage. The 85/1.4L IS has generally been a lot better and the 35 IS is also good, though not necessarily great.
...
That is not a good feeling to get from using the trusted lenses. Esepcially if the sensor really outresolves the lens. If it is mostly reduced actuance at the pixle level, I see that as a good thing as the sensor resolution is then optimal for resolving most detail and preventing aliasing.
That being said, after some years using mostly zooms for landscape, as they became really good some years ago, to the extent that I didn't see any point in using my classic Zeiss primes when used stopped down, I am planning to use prime lenses more again. The TS-E 135 macro will greatly outperform any 70-200 zoom (check infinity MTF at TDP), and for 35 mm I will use the Milvus 1.4 which is even stronger than the Canon L II in the outer frame stopped down moderately.
Camperjim wrote:
At ISO 6400 you can obtain results that are close to ISO100.
I am curious about what you mean with "close". If I use DPP (latest version) to reduce noise at ISO 6400 to natural noise level at ISO 400, the detail is muddy and the finest detail is gone. Nothing like ISO 100 at all (at 100% magnification).
rscheffler wrote:
I get what you're saying and one aspect of 32MP that appeals to me is the amount of detail that carries over when downscaling vs. images shot natively at that resolution. That said, when pixel peeping, it's a bit disconcerting that a decent FF lens is showing problems within the area of APS-C coverage. The 85/1.4L IS has generally been a lot better and the 35 IS is also good, though not necessarily great.
...
alundeb wrote:
That is not a good feeling to get from using the trusted lenses. Esepcially if the sensor really outresolves the lens. If it is mostly reduced actuance at the pixle level, I see that as a good thing as the sensor resolution is then optimal for resolving most detail and preventing aliasing.
That being said, after some years using mostly zooms for landscape, as they became really good some years ago, to the extent that I didn't see any point in using my classic Zeiss primes when used stopped down, I am planning to use prime lenses more again. The TS-E 135 macro will greatly outperform any 70-200 zoom (check infinity MTF at TDP), and for 35 mm I will use the Milvus 1.4 which is even stronger than the Canon L II in the outer frame stopped down moderately. ...Show more →
The problem I noticed in some 70-200/4L IS images, shot at infinity, was uneven corner sharpness/softness, with some cases showing unacceptable softness. But it wasn't consistently visible in all images. I haven't used the combo enough, but it likely changes with focal length and/or subject distance. Some nearer distance subjects that were roughly planar showed consistent across-frame sharpness. Another aspect is shooting technique. Given it was a hike with photography, rather than a hike for photography, technique wasn't necessarily perfect. Hand-held at low ISOs (in decent light) at f/5.6, I relied on IS to offset the somewhat marginal shutter speeds for this level of sensor pixel density.
I have a mid-70s vintage Leica 180/3.4 Apo Telyt that sharper and more consistent across the frame than the 70-200/4L IS... Maybe I should look at Leitax adapting it, though realistically I will want to use it in live view and a mirrorless camera would be a better solution, ideally one with IBIS. Which of course currently rules out any Canon cameras. Or just bring a tripod along, though that changes the dynamics of some outings.
Thanks for the suggestions to try DPP again... Maybe I should give it another try, but I will not return to it for 'production' where workflow efficiency is desired. It also won't work with processing my non-Canon raw files. I guess the roundabout question I had was: for those processing 90D files with Adobe software, are you happy with the results? Do you find you need to tweak the default settings much? Is the Adobe profile the one you use, or do you prefer one of their emulations of Canon's picture style? For example, when processing 1DXII files, I dislike the Adobe profile most of the time and use their emulation of Canon's 'natural' picture style, though with 1DX files, I'm happier with Adobe's profile and usually use it.
Maybe you need to send the 70-200 4L IS in for alignment. I owned the prior model (non-IS) and it started as a really sharp lens but drifted out of alignment, probably from carrying it on a bike and/or running with it. One side was softer and corners were out. It was out of warranty but for some reason Canon fixed it for free. The 70-200 4L IS I replaced it with is still tack sharp after 9 years but I'm sure given the right conditions can drift out as well. If the corners are soft on APS-C something is wrong.
It could be... though corner softness is not consistent across the zoom range, which I understand to be typical behaviour of zooms. I should also do some tests that are better controlled to hopefully determine where the problem lies.
Gochugogi wrote:
Maybe you need to send the 70-200 4L IS in for alignment. I owned the prior model (non-IS) and it started as a really sharp lens but drifted out of alignment, probably from carrying it on a bike and/or running with it. One side was softer and corners were out. It was out of warranty but for some reason Canon fixed it for free. The 70-200 4L IS I replaced it with is still tack sharp after 9 years but I'm sure given the right conditions can drift out as well. If the corners are soft on APS-C something is wrong....Show more →
I have a similar problem as rscheffler with the 55-250 STM that seems related to the IS or focus mechanism: for some focal lengths, especially around 150-200mm, there is very uneven sharpness across the frame SOME of the time, even for near-infinity subjects. One frame can be fine and the next one with same settings not acceptable so it is not a simple alignment problem. The problem would not be visible on the LCD or when posting web size images of maybe 1 MPixel. But with the 80D when viewed on a good monitor, sometimes about 1/3 of the frame has a very visible blurry look that makes it unfit for quality images. The problem seems to occur more frequently with fast shutter speeds which makes me suspect the IS unit (in combination with shutter/mirror shock) as the source of the problem.
At the same time, at most focal lengths including the ends of the range no problem, pretty even sharpness across the frame with the typical loss of corner sharpness that is to be expected. It's not due to wear-and-tear either, as my lens has been like that from the start (but took some time before I noticed, and with 80D far more obvious than with the 450D before that).
Very interesting post. It's like you were sitting here with me looking at my photos and describing the problem for me. I just went through the first portion of the images I shot with the 70-200/4L IS v1, which were primarily infinity scenes: a forest viewed from an elevated position.
Images from the extremes of the zoom range generally looked great. Some, but not all, images in the middle to higher end showed uneven sharpness. In my case, usually the left side was worse than the right. For example, an image made at 144mm showed uneven sharpness, but one made moments later at 165mm looked good. During this phase of my shooting, I was somewhat lazy and had the camera set in aperture priority and auto ISO with the minimum shutter speed weighting set to +2. Exposures were in the 1/500-1/1250 range at f/5.6 with ISO floating around between 100 and 400. Just came across an image shot at 89mm where a tree in the relative foreground front left was sharp and the forest farther away on the right side was sharp, implying a tilted plane of focus.
There were also a couple photos where the camera's AF missed slightly compared to similar images before and after those moments, which also revealed a greater tendency to more softness on one side than the other.
It seems to me that there are a number of factors at play: my technique, IS, AF/focus...
alundeb wrote:
I am curious about what you mean with "close". If I use DPP (latest version) to reduce noise at ISO 6400 to natural noise level at ISO 400, the detail is muddy and the finest detail is gone. Nothing like ISO 100 at all (at 100% magnification).
I am not sure why the difference. I allow DPP to select the processing for opening c-raw files. The ISO6400 will then need a touch more contrast, saturation, and sharpening. Then results will look very similar and prints up to my Epson 3880 printer max of 16x24 will look virtually identical. I get pretty much the same with out of camera jpegs. Certainly jpegs at ISO3200 should look clean.
Camperjim wrote:
I am not sure why the difference. I allow DPP to select the processing for opening c-raw files. The ISO6400 will then need a touch more contrast, saturation, and sharpening. Then results will look very similar and prints up to my Epson 3880 printer max of 16x24 will look virtually identical. I get pretty much the same with out of camera jpegs. Certainly jpegs at ISO3200 should look clean.
I could probably print ISO 3200 from the 90D at 16x24, with NR turned completely off, and it would look great. Still at close inspection after default NR it looks clearly inferior to ISO 100. Anecdotal image quality comparisons usually come out wildly different just because they are highly inaccurate by nature.
rscheffler wrote:
I have a mid-70s vintage Leica 180/3.4 Apo Telyt that sharper and more consistent across the frame than the 70-200/4L IS... Maybe I should look at Leitax adapting it, though realistically I will want to use it in live view and a mirrorless camera would be a better solution, ideally one with IBIS. Which of course currently rules out any Canon cameras. Or just bring a tripod along, though that changes the dynamics of some outings.
The Leica 180/3.4 Apo Telyt is a sweet lens. I almost regret selling mine to a fellow who adapts it to medium format (GFX). It is not as sharp in the center as some other modern lenses, but the evenness of the field more than makes up for that. The correction of axial color is also really good.
Problems with consistency of zoom lenses with IS makes another argument for IBIS. I follow the advice from Roger Cicala, don't commit to a mirrorless mount in 2019. There are too many things in the pipe that we don't know right now (That was before the A7r IV). By the mid of 2020 I think we will know what Canon has in mind for a high resolution EOS R camera, and if it will have IBIS. Of course this doesn't help 90D users.
alundeb wrote:
Problems with consistency of zoom lenses with IS makes another argument for IBIS. I follow the advice from Roger Cicala, don't commit to a mirrorless mount in 2019. There are too many things in the pipe that we don't know right now (That was before the A7r IV). By the mid of 2020 I think we will know what Canon has in mind for a high resolution EOS R camera, and if it will have IBIS. Of course this doesn't help 90D users.
But IBIS doesn't work well with longer lenses ... (and probably same story for macro lenses). But I agree that for more general use at short-medium focal length and medium to long shooting distance, IBIS is better once the problems of the technology (heat dissipation, power, vulnerability etc.) are ironed out.
Not a good time indeed to commit to a certain mount, we don't know what is going to happen with ML AF technology (no sign yet that the Sony A9 approach will catch on outside a niche market and no real alternative in sight for fast action), with APS-C format (Canon direction unclear, Nikon introduced an APS-C ML camera but no sign of serious APS-C lenses just like with DSLRs) and with computational imaging in general (where ML cameras are still seriously lagging smartphones with the difference probably getting even bigger).
With a continuing decline in sales for the next years there is a big risk that the ML market will see more incremental change just like the DSLR market lately, which in addition to higher prices could start snowballing in a negative way.
120 fps clip.
Some people might be surprised at getting this quality footage from a camera you can buy now for $1099.99.
Shot from 300 meters away on shore. Action on an island. 800mm + 1.4x vII + 90d.
Andrew J wrote:
120 fps clip.
Some people might be surprised at getting this quality footage from a camera you can buy now for $1099.99. From a Canon Authorized dealer, no shipping fees, and no tax(unless you are in that dealers state). More info at canonpricewatch.com
Shot from 300 meters away on shore. Action on an island. 800mm + 1.4x vII + 90d.
impressive coming from canon 90D! thanks for the effort and the sharing. thumbs up ;-)
