RustyBug wrote:
Well, as you know ... I'm tapped at the 100-400 II + 1.4X III on the long end. How it rolls on the AF with the f/8 combo (native 5.6 also) is of interest. That, and if the buffer hangs you up if you try to resume shooting after a burst that is still buffering.
BTW ... I "liked" your post, but really wanted to "Dislike" it out of jealousy for Rob's camera.
Mostly, just put it through your own paces, and I'll watch & listen vicariously for a bit.
Very brief play time with it tonight before the sun went down but here are a few things I noted.
Using the same Card Andrew mentioned earlier in this thread.
SanDisk 64GB Extreme PRO SDXC Memory Card, UHS-II @ 300MB/s Read
Shooting full RAW
1/160 ss, ISO 640. AI Servo mode
OVF
Solid 10 fps - verified by time stamps on EXIF data on several different burst.
Filled buffer after 29 frames. If you continue to hold the shutter down it would fire another 3 images after about a 1 second pause then pause again about 1 second then it fired off another 5 or so frames. Not sure what it does if you stay right on it beyond that.
I got about 38 frames total in 4.5 seconds before I stopped which includes the few sputter frames at the end.
The buffer does not block you from firing off another burst after letting off the shutter but it seems to limit the amount of frames to however many have cleared. For instance, I let off the shutter after hitting the buffer then fired it again after about 1 second and got about a 3 frame burst. I then waited about 2 seconds and got another 6 or so frames.
Again this was with full raws and ran out of daylight before could try cRaw.
Shooting with the 100-400II AF seems pretty snappy to me. It seemed faster then the 6DII. I added the 1.4x to the combo and I could see it did slow down a little and of course as the light was fading it was taking a larger hit, but at f/8 I think that's to be expected. Again, I was pressed for daylight/time so I need to evaluate this much more when time permits.
I didn't even have time to slap the 500II on her tonight.
Bottom line is I can already see a replacement shutter in my future if I don't learn to control myself with this little beast. I love that machine gun sound. I don't know how only 4 fps over my 6DII @ 6 fps can be so drastically different but it is VERY noticable and VERY welcomed. I can't wait to get out there to do some real shooting with it.
Oh, one other pleasant surprise was to see how good the sooc raw images looked after uploading them and viewing them in LR.
Here's a quick ISO 6400 shot BEFORE (no sharpening or NR applied) and an AFTER (cleaned up)
Andrew J wrote:
It can be fixed by tuning the highly customizable AF system, which they refuse to do. That is fine but they don't even say "we refused to adjust the most basic tracking settings for our conditions."
Sharlin wrote:
Well, the DPR guys have weird aversion aversion to the Custom Functions menu, and given that the AF settings are there, they're probably pretending they don't exist at all.
They also claim to hardly use the AF joystick, which seems rather bizarre for any serious use. I wonder if they calibrate AF micro adjust per lens. If not, that could explain OOF images.
lighthound wrote:
The buffer does not block you from firing off another burst after letting off the shutter but it seems to limit the amount of frames to however many have cleared. For instance, I let off the shutter after hitting the buffer then fired it again after about 1 second and got about a 3 frame burst. I then waited about 2 seconds and got another 6 or so frames.
Good to know.
I love that machine gun sound.
Reminds me of the sound my 1D2 N used to make at 8 FPS. If it has that similar cadence ... yup, it can be rather visceral. It gives you that feeling of "Now, that's a camera."
Since I haven't shot it yet, I can't compare it directly, but your statement sure is reminiscent of when I got my first 1D2.
I don't know how only 4 fps over my 6DII @ 6 fps can be so drastically different
6 fps = .166 sec
10 fps = .1 sec
So while it is "only" 4, it takes 66% more time between shots on the slower 6D2. Which, essentially means for every 2 shots you got with the 6D2, you now get 3 1/3.
Took a stab at your sooc (usm & nr only, global applied) to see how it played a bit (this was using CS6, i.e. old laptop). Will be a learning curve to find out what the files can tolerate and how to best harness them through what settings / approach to get that sweet spot between noise & detail. First crack (from the posted small jpg) is encouraging.
From your exif, that calc's to about EV 10 lighting.
Next up ... 500/4 at dawn with a K-Mart, blue light special.
NOTE: Since the Bayer array is RGGB or RGBG, there is only 1 B, so the same ISO 6400 with full on, deeper blue light may be a bit noisier than presented here. Will be interesting to find out how it plays for ya. But, if f/8 is netting this, I'm sure you're looking forward to f/4, all the more.
RustyBug wrote:
Reminds me of the sound my 1D2 N used to make at 8 FPS. If it has that similar cadence ... yup, it can be rather visceral. It gives you that feeling of "Now, that's a camera."
It's not so loud as that. It's just a faster version of the 70D/80D.
If you wanted visceral, try the 1vHS. At 10FPS everyone could hear it from a mile away.
RustyBug wrote:
They haven't posted up the ISO invariance yet for the 90D. While the 80D isn't ISO invariant as much as say the D500, it will still be interesting to see if the 90D improves (or just matches) in the low ISO arena as well.
This is looking really good, Rusty. I prefer looking at the expousre latitude rather than ISO invariance, and here I think it is better than the 80D and now right up there with the latest Sony. The effect of higher resolution on low ISO dynamic range that exists in theory, also shows in the images.
Here is 6 stops push, rather extreme. Notice the relative lack of color cast in the 90D shadows. I really can't wait to get those pixels in a FF sensor
alundeb wrote:
This is looking really good, Rusty. I prefer looking at the expousre latitude rather than ISO invariance, and here I think it is better than the 80D and now right up there with the latest Sony. The effect of higher resolution on low ISO dynamic range that exists in theory, also shows in the images.
Here is 6 stops push, rather extreme. Notice the relative lack of color cast in the 90D shadows. I really can't wait to get those pixels in a FF sensor
However in that example, as you move the viewport around, like over the hair extension, the 80D has more detail, the 90D loses a bit of the definition. Not sure if it just their comparison, or if it is the gear, but in that situation, you trade color blotching for detail.
If this is true, it leads one to believe that Canon has introduced some color noise reduction into the raw file itself, thus making it look better than before, but at a slight loss of detail, and thus any such improvements aren't really sensor, but software-produced. That wouldn't surprise me, especially if Canon has pulled all tricks at the hardware level they could, and now they have to manipulate the raw. Other manufacturers do this as well, some to greater extents than others. Canon has always been very conservative when it comes to raw data in the past.
Based on the test images from DPR's various tests, the new sensor is 'slightly' cleaner at high ISO than the 80D was, not a big difference. What is different, as noted by Alundeb, is that the recovery is also cleaner and loses what looks like a magenta cast (similar with the Sony a6400) when pushed 4 or more stops. The overall "push-ability" in terms of basic noise is similar, whereby you'll see about the 3.5 stop limit, but going a little farther doesn't result in unwanted coloration.
TeamSpeed wrote:
However in that example, as you move the viewport around, like over the hair extension, the 80D has more detail, the 90D loses a bit of the definition. Not sure if it just their comparison, or if it is the gear, but in that situation, you trade color blotching for detail.
If this is true, it leads one to believe that Canon has introduced some color noise reduction into the raw file itself, thus making it look better than before, but at a slight loss of detail, and thus any such improvements aren't really sensor, but software-produced. That wouldn't surprise me, especially if Canon has pulled all tricks at the hardware level they could, and now they have to manipulate the raw. Other manufacturers do this as well, some to greater extents than others. Canon has always been very conservative when it comes to raw data in the past....Show more →
The image is a combination of sensor data and software that manipulates that data. Personally I don't try to distinguish. I am only interested in the final outcomes. Software is not a trick. It is a necessity that is required to process sensor data.
Trying to decide the difference between sensor performance and software manipulation is pointless. There is no way to "see" a raw image. When we attempt to look at a raw file, what we see on our monitors is a highly manipulated output.
When you compare raws from an 80D, or a 90D, or an image from another brand of camera you are looking at the manipulated data. With the DPR studies we are looking at the output from Adobe ACR which compared with DPP seems to do a poor job with the 90D data. Maybe ACR will catch up but for now if you want to see the best results you will need to use DPP.
I can GUARANTEE you that if I have unadulterated raw data, ie. no noise reduction algorithms run on the raw at all, I have access to tools that can clean up that data better than a specific algorithm that Canon chose to do this on the raw that can fit in the firmware update.
So indeed, I am in the camp of believers that I don't want Canon "baking" the raw files, I want it to be as unadulterated as possible, because I can get better results than Canon can. They only have so much room for the firmware, and can only choose a set algorithm, given the ISO levels and NR levels.
We both can be satisfied if and only if Canon would actually have a NR setting that was beyond "Disabled", which isn't really disabled, but just a reduced setting. If I had the option of "Not Applied", for example, Canon could leave everything alone, the raw and JPG, presuming there are now changes in the raw processor that wasn't there to the degree it is in the 90D. You, and others like you, that are happy with the Canon output after processing could then use Disabled, Low, Standard, whatever for that setting and then can just accept the results as they come out of the software.
Yes, the differences on DPreview could just be the tools used, I already stated that, and I only offered the observation as I saw it. I always use DPP as it is, I never use Adobe tools for raw conversions. Canon nails things out of the gate for "good enough", Adobe has to play catchup. However, I am very curious that if I disable the NR setting in DPP, for certain files like the 90D, there may already be something going on with the raw. I hope not...
Mabye, but the difference between custom processing from unaltered raw and custom processing from weak color NR raw, in this case, ISO 100 pushed 6 stops, would be so small that it has no practical consequence.
BTW I find CaptureOne to be closer to DPP in basic demosaic than ACR.
Its good we can push it that far, but until we have a truly invariant ISO camera, folks will raise their ISOs up to a level that produces a decent exposure and then only really push shadows 2, possibly 3 stops. I have had no issues with the 7D2 doing that, or the M50 quite frankly, so the real-world differences are very, very small with the 90D to its siblings/big brother, I feel.
I suspect Canon is keeping its latest sensor tech under lock and key until it can be presented on a mirrorless body, it won't see implementation on any remaining DSLRs, however few there will be in the future.
TeamSpeed wrote:
I can GUARANTEE you that if I have unadulterated raw data, ie. no noise reduction algorithms run on the raw at all, I have access to tools that can clean up that data better than a specific algorithm that Canon chose to do this on the raw that can fit in the firmware update.
So indeed, I am in the camp of believers that I don't want Canon "baking" the raw files, I want it to be as unadulterated as possible, because I can get better results than Canon can. They only have so much room for the firmware, and can only choose a set algorithm, given the ISO levels and NR levels.
We both can be satisfied if and only if Canon would actually have a NR setting that was beyond "Disabled", which isn't really disabled, but just a reduced setting. If I had the option of "Not Applied", for example, Canon could leave everything alone, the raw and JPG, presuming there are now changes in the raw processor that wasn't there to the degree it is in the 90D. You, and others like you, that are happy with the Canon output after processing could then use Disabled, Low, Standard, whatever for that setting and then can just accept the results as they come out of the software.
Yes, the differences on DPreview could just be the tools used, I already stated that, and I only offered the observation as I saw it. I always use DPP as it is, I never use Adobe tools for raw conversions. Canon nails things out of the gate for "good enough", Adobe has to play catchup. However, I am very curious that if I disable the NR setting in DPP, for certain files like the 90D, there may already be something going on with the raw. I hope not......Show more →
Looking at the file from Dave's Leaves (see "Guess The ISO" thread) ... I didn't get any kind of feeling that they were pre-baked ... particularly after turning the default NR back to "0"
TeamSpeed wrote:
I can GUARANTEE you that if I have unadulterated raw data, ie. no noise reduction algorithms run on the raw at all, I have access to tools that can clean up that data better than a specific algorithm that Canon chose to do this on the raw that can fit in the firmware update.
So indeed, I am in the camp of believers that I don't want Canon "baking" the raw files, I want it to be as unadulterated as possible, because I can get better results than Canon can. They only have so much room for the firmware, and can only choose a set algorithm, given the ISO levels and NR levels.
We both can be satisfied if and only if Canon would actually have a NR setting that was beyond "Disabled", which isn't really disabled, but just a reduced setting. If I had the option of "Not Applied", for example, Canon could leave everything alone, the raw and JPG, presuming there are now changes in the raw processor that wasn't there to the degree it is in the 90D. You, and others like you, that are happy with the Canon output after processing could then use Disabled, Low, Standard, whatever for that setting and then can just accept the results as they come out of the software.
Yes, the differences on DPreview could just be the tools used, I already stated that, and I only offered the observation as I saw it. I always use DPP as it is, I never use Adobe tools for raw conversions. Canon nails things out of the gate for "good enough", Adobe has to play catchup. However, I am very curious that if I disable the NR setting in DPP, for certain files like the 90D, there may already be something going on with the raw. I hope not... ...Show more → RustyBug wrote:
Looking at the file from Dave's Leaves (see "Guess The ISO" thread) ... I didn't get any kind of feeling that they were pre-baked ... particularly after turning the default NR back to "0"
Shouldn't you have the 90D by now, even if Smartpost or another slow shipping method was selected?
I don't know how only 4 fps over my 6DII @ 6 fps can be so drastically different but it is VERY noticable and VERY welcomed. I can't wait to get out there to do some real shooting with it. 
