70D ISO Test, Straight from the Camera, JPG's

Paul Iddon

OK, testing the ISO levels on the new 70D. The lens is my EF 100mm f/2.8 macro.

These were taken tripod mounted, all at f/11 in AV priority. The camera was set to it's maximum JPG setting and the results saved and displayed as full size images.

Opened into Photoshop, no adjustments whatsoever and uploaded to Photobucket (if you see softness, then that's going to be due to Photobuckets compression routines. The only editing is where I added text to show the ISO numbers.


















Paul.

insanefred

Were those photos take out doors with a over cast sky; good lighting? High ISO tests with good lighting are not worth much now days. Try again with really bad lighting.

pathfinder

5 stops faster than Tri-X and folks still complain....Go figure.

Thanks for uploading these images, Paul.

I think these look remarkable for a crop sensor camera. I notice the decrease in contrast and color density much more than grain in the 6400 and 12800 ISO images.

Images obtained under "really bad lighting" - not sure what is being requested here.... Harsh, high contrast sunlight is bad lighting to me.

insanefred

pathfinder wrote: »

5 stops faster than Tri-X and folks still complain....Go figure.

Thanks for uploading these images, Paul.

I think these look remarkable for a crop sensor camera. I notice the decrease in contrast and color density much more than grain in the 6400 and 12800 ISO images.

Images obtained under "really bad lighting" - not sure what is being requested here.... Harsh, high contrast sunlight is bad lighting to me.

I didn't ever mention under harsh direct sunlight. Perhaps I need to be more specific.
Try some night life photos; think photojournalism. Capture some action in low light, let's just say a inside a gym (basketball, MMA, etc...) Just something realistic.
As comparing with film, print or slide. It is getting really old, this is a digital forum after all. (yes, I know it still welcomes film, and film talk)
Again, who really thinks ISO tests are even remotely useful in good lighting other than manufactures? Unless you are that guy that keeps forgetting to change from ISO 6400 back to 200 when it is optimal.

Matthew Saville

BS!!!!

(Keep in mind that I say this jokingly, please don't get all offended that I would question your integrity or whatever other silly things people get all personal and aggressive about...)

But honestly, that test looks insanely impossible. It looks like you used the same exact image for ISO 100-1600. And if that is what a crop sensor can do at ISO 12800, well damn, full-frame lovers had better start shutting up! I don't think the D800 / 5D mk3 look this good at ISO 12800, and maybe not even the D4 / 1DX. This must be some insane quantum leap in sensor technology, because the last crop sensor I tested, that (your 12800 sample) is what ISO 1600/3200 looked like...

Of course as insanefred mentions, even if this is totally legit the REAL test is how this sensor performs in low-light conditions, with deep dark shadows and 1-2 stops of under-exposure, and of course, RAW not JPG.

=Matt=

JC

Did Canon fix the sensor corduroy in the lowlights? I shot simultaneously all night for the perseids with my 5D mk II and OM-D E-M5, and while the larger sensor has, overall, better light gathering ability, the sensor on the E-M5 is amazing. No matter how much I stretch the lower end of the histrogram, the noise is completely random, except for the bayer filter pattern that shows up at about 800% magnification, and a raw noise correction can work on that. The corduroy in the 5D is much worse by comparison, and raw noise correction doesn't work on it.

Paul Iddon

insanefred wrote: »

Were those photos take out doors with a over cast sky; good lighting? High ISO tests with good lighting are not worth much now days. Try again with really bad lighting.

Fairly overcast - no direct sunlight, taken on top of the wheelie bin in the garden!


Paul.

perroneford

I am confused as to what these images are supposed to show. What was being tested here?

JC

perroneford wrote: »

I am confused as to what these images are supposed to show. What was being tested here?

the increase in noise (or lack thereof ) at progressively higher ISO, for the same level of camera-automatic exposure.

Paul Iddon

Matthew Saville wrote: »

even if this is totally legit

=Matt=

These jpg's really are straight out of the camera Matthew...

Exif is intact and included for examination - feel free to download and check if your browser doesn't allow you to check directly m8.

perroneford

kolibri wrote: »

the increase in noise (or lack thereof ) at progressively higher ISO, for the same level of camera-automatic exposure.

Well ok, in that instance this test is fundamentally flawed. Noise primarily exists in the shadows of an image. Not in the well exposed areas. By essentially eliminating the shadow areas of the image very little noise will be in evidence.

The test would be FAR more realistic in shooting a stairstep chart or something similar. Even as a practical matter, one would hardly use ISO 12,800 in a brightly lit environment unless extreme shutter speeds were needed. A more practical test would be to light a scene, increase the shutter speed until the desired ISO was *mandated* for a proper exposure, and then see what that looks like. For instance a room lit by a single light fixture, starting at ISO 100 and maybe a shutter speed of 1/60th or whatever is appropriate. Then increase the shutter speed by full stops until an ISO of 12,800 is called for.

puzzledpaul

Paul - should you get the chance, I'd also be quite interested to see some low light / murky hi iso shots - but real world, outside stuff, tracking subjects like birds etc.
Whilst I can imagine the weddings / pj etc mob being interested in candle lit scenarios, I don't come across too many of those @ my usual venues.

I appreciate it'll be a difficult task - waiting for a murky day - but ... thought I'd ask anyway ...

How's the AF in tracking / servo situations?

pp

perroneford

Well the REAL scenario for me, is shooting in gymnasiums, auditoriums, etc. That's a real world scenario that I see as a sports shooter, and I'd imagine many would see shooting indoor performances of all kinds. Then there is the overall darker scenario of shooting high speed subjects in marginal light, such as twilight birding or similar.

paddler4

I never shoot jpeg, but don't canon cameras apply NR to JPEGs? If so, what I would find more telling would be raw images, with no NR and no other editing other than the initial rendering. I agree that shadows are where the problem is worst, so an image with high tonal range would be helpful (although my camera shows ample noise at very high ISOs throughout the image). Paul, if you get even approximately similar results in raw files near the left tail of the histogram, I will be very impressed (and thinking harder about replacing my current camera).

The discussion of lighting puzzles me. Within a reasonable range of exposure times, two images shot under different amounts of light, but with offsetting aperture differences, should behave the same. E.g., if one quadrupled the amount of light but closed the aperture by two stops, the amount of light hitting the sensor would be identical. As long as the tonal range of the image extends all the way down to 0,0,0 or close, the noise should be similar. Am I missing something here? I'm not the slightest interested in high ISO performance in bright lighting, but I am very interested in high ISO performance in lower light conditions, and that that should be replicable under brighter light.

Night photography is an exception to this, for two reasons. One is that very long exposures produce a lot of pattern noise. That's why Canon offers long-exposure (subtractive) noise reduction, which does not obscure detail and in my experience is quite effective in reducing pattern noise. A second reason is that sensors heat up with long exposures, particularly in warm weather, increasing both random noise and other ugly stuff. E.g., with long enough exposures at summer temperatures, mine develops irregular color distortions. So, I would want a separate set of tests for really long exposures. This is why some serious night photographers still keep a film body--because even the digital sensors most resistant to overheating simply can't manage the extremely long exposures that you can get with film

WilliamClark77

These are pretty much worst case scenario: Sunset, shaded, and underexposed. Shot in raw, opened in DPP (Lightroom can't open the raw files yet), no sliders touched except adding +1/3 exposure, auto lighting optimizer off. Raw sharpness was at +7. I would've turned that down had I noticed before uploading. 70-200F4L lens at f5.6. All shots straight out of dpp saved at a quality of 9 (1-10 scale) except the ISO12,800 shot. I opened it in Photoshop and saved at a quality of 10 (1-12 scale) to reduce the file size from 23.5mb to 11.5mb.

ISO1600
1/50 shutter speed



100% crop



ISO3200
1/160 shutter speed



100% crop



ISO6400
1/200 shutter speed



100% crop



ISO12,800
1/320 shutter speed



100% crop

JimKarczewski

Just for comparison, I have some 1Dx photos in crappy light.. Because a lot of High Schools have crappy light. So this is a 1Dx at 12,800

1x1 crop (sorry, newspaper thing)


100% Detail Crop


No noise reduction, color correction and clarity adjustments. Unfortunately this is from a RAW file, but I barely touch the files. There is some blue Chromatic abberation due to the use of a 1.4x TCIII on the 70-200/2.8 I

ISO16,000 from tonight, again 1Dx with 70-200 & 1.4TCIII, 100% Crop


Original File

Paul Iddon

The two 12,800 compare pretty well considering the specification of the two cameras and the FF vs crop sensor.


Paul.

divamum

Paul Iddon wrote: »

The two 12,800 compare pretty well considering the specification of the two cameras and the FF vs crop sensor.


Paul.

Agreed.

I do have to wonder if the 70d IS in fact the "7d II". Other than build and weathersealing I can't for the life of me imagine what any 7dII will have that this doesn't (or maybe Canon has something HUGE up their sleeves for that - in which case, I'm prepared to be gobsmacked!).

If I were looking for a crop camera, I'd get the 70d without hesitation - the spec is great, and the samples I've seen look outstanding. It's clear that they have moved the xxd series back up in the line; the 60d seemed to be kind of a "Rebel +", where the 70d has an awful lot more to offer. Canon has a winner here, I think, particularly at the price point it has launched.

puzzledpaul

divamum wrote: »

.... Other than build and weathersealing I can't for the life of me imagine what any 7dII will have that this doesn't ...

+ An AF point layout based on 1D / 5Dm3 layouts ... I'd hope for, rather than any diamond shape, irrespective of how many points it's filled with

pp

divamum

Dunno about that - my 7d's 19pt af layout has never left me lacking. As long as I have useable AF points in the rule-of-thirds intersections so I can focus on EYES I'm happy!

insanefred

cat photos you posted, look like they been over sharpened a tad. Otherwise they look decent.

jmphotocraft

divamum wrote: »

I do have to wonder if the 70d IS in fact the "7d II". Other than build and weathersealing I can't for the life of me imagine what any 7dII will have that this doesn't (or maybe Canon has something HUGE up their sleeves for that - in which case, I'm prepared to be gobsmacked!).

A 7D2 could have 10fps and an upgraded AF system, like the one in the 5D3 perhaps, and sell for $1899 or $1999 as the premium APS-C camera. It would be the "affordable" semi-pro sports camera. I think this would be a very logical move for Canon, and if they don't do it I'll be surprised and disappointed. In fact I'd be tempted to sell my 5D3 and 300/2.8 to get it, if I didn't think that the sight of the 300/2.8 itself at sports events helps generate sales.

However Canon could be waiting to respond to the presumed Nikon D400.

perroneford

jmphotocraft wrote: »

However Canon could be waiting to respond to the presumed Nikon D400.

I don't think either one is waiting. I think neither are ready to release a product. The time to have done it would have been about 6-8 weeks ago if they were targeting pro sports shooters. Releasing a new camera for NFL preseason is where you want to be. If they were trying to hit basketball season, we would be seeing something now. Basketball is only a couple months off, and no one wants to take a new, untested camera into the real season.

Nikon just need to give me the D600 sensor in a pro body, call it the D4s and TAKE MY MONEY. A 7D2 at $2k would be a hard price point since Nikon's entry level full frame camera is there, and their flagship crop body is about $800 cheaper.

JimKarczewski

If you are a CPS member the 60D (and beyond) won't qualify for any points toward membership. They will only accept points from the 1D, 5D, 6D, and 7D products..

http://www.cps.usa.canon.com/misc/cps_program_notice.html

They will repair them if you are a CPS member but, won't qualify you for different levels...

Paul Iddon

I don't see how any company would release a model in time for a specific sports genre. There is no NFL in the UK, we have soccer as our national sport and I'm sure Canon UK wouldn't guide Canon to release a model in time for the new season or the Champions League starting up in August/September. Release dates by camera manufacturers is when they want them to be, not to coincide with an American sporting event. Unless you know for sure otherwise of course?

Well done on getting published btw. That's always a great thing to have in your portfolio.

Paul.

jmphotocraft

perroneford wrote: »

Nikon just need to give me the D600 sensor in a pro body, call it the D4s and TAKE MY MONEY.

Indeed that would be a sweet camera.

A 7D2 at $2k would be a hard price point since Nikon's entry level full frame camera is there, and their flagship crop body is about $800 cheaper.

$1899-2000 would be easy if it has 5D3 AF and 10fps. Canon's entry level full frame camera is also at that price point. That's fine. With the D600 and 6D, you are paying for the FF sensor and accepting a compromise on everything else. With my imaginary 7D2, you'd be paying for the AF and 10fps, and accepting a (now even smaller) compromise on the sensor. Two different cameras for different purposes at the same price point. Shouldn't be a problem.

And Nikon's flagship crop body doesn't really exist right now. A D400 will surely be introduced at a similar price and spec to the 7D2. The 7D was $1700 at first.

perroneford

On the surface, I'd say you are right. Camera manufacturers seem to get at least working prototypes of bodies and lenses in certain people hands at major events such as the World Cup, the Olympic Games, the Superbowl, etc.

I just looked at the release dates for some of Nikon's most prominent cameras and they do seem to fall all over the map in terms of year, but not timing of year:

January: D4
February: D2HS, D1x, D1H, D800,
March: None
April: D800E
May: None
June: D2XS, D1
July: D2H, D300S, D700
August: D3, D300
September: D600, D2X
October: D3s
November: D200
December: D3X


by year:

1999 D1
2000 None
2001 D1X, D1H
2002 D100
2003 D2H
2004 D2X
2005 D2HS, D200
2006 D2XS
2007 D3, D300
2008 D3X, D700
2009 D3s, D300s
2010 None
2011 None
2012 D4, D800, D800E, D600

I'll leave correlation to sporting events to others who know better than me.

Paul Iddon wrote: »

I don't see how any company would release a model in time for a specific sports genre. There is no NFL in the UK, we have soccer as our national sport and I'm sure Canon UK wouldn't guide Canon to release a model in time for the new season or the Champions League starting up in August/September. Release dates by camera manufacturers is when they want them to be, not to coincide with an American sporting event. Unless you know for sure otherwise of course?
Paul.

Paul Iddon

This was taken in the garden, at ISO 1600 on the tripod, so a long exposure. Only edit was a burn across the brightest area a bit, but only a small amount. No sharpening, or NR applied, no other editing at all. Transferred from RAW to a tiff, and saved out as jpg.

I saved it as a jpg at size 3 for web uploading purposes and re-sized on Imageshack to 1024px. The light is the outside security light on the wall of the house, 5 or 6 yards away.


Exif:

Camera Maker: Canon
Camera Model: Canon EOS 70D
Lens: EF100mm f/2.8 Macro USM
Image Date: 2013-09-11 21:15:38 +0000
Focal Length: 100mm
Aperture: f/10.0
Exposure Time: 3.200 s
ISO equiv: 1600
Exposure Bias: none
Metering Mode: Matrix
Exposure: aperture priority (semi-auto)
White Balance: Auto
Flash Fired: No (enforced)
Orientation: Normal
Color Space: sRGB
GPS Coordinate: undefined, undefined
Creator: Paul Iddon
Copyright: Paul Iddon - A View of the UK







Paul.

WilliamClark77

insanefred wrote: »

cat photos you posted, look like they been over sharpened a tad. Otherwise they look decent.

Yup. I turned the raw sharpness all the way up to see any noise better and forgot to turn it down to a sensible level before exporting to jpeg and uploading. Those shots weren't meant to win any prizes, just a scene with a lot of dynamic range to test on.

I have been more impressed with this camera than any other I've used. Even more impressed than when I got my first dslr, a Rebel, back when they first came out and had nothing to compare it to but point and shoot cameras.

I'm really impressed with the high iso capabilities, but that's just the tip. The menu layout and touch screen is very easy to use and intuitive. Even having never used this body, most everything is located where you would expect it to be and not buried in some senseless subcategory that makes you look like you're texting someone directions instead of shooting. The autofocus is quick and responsive, even in live view. As in, it focuses just as fast and tracks just as good as looking through the viewfinder. I've used it very little though to really test it out.


Also, I've been quite impressed with the kit 18-135 stm lense. I only wanted the body, but all that was available was the kit with it. From the very few pictures I've taken, it competes quite closely with my L glass as far as image iq is concerned, not on build quality though and a constant wide aperture would be nice. It focuses fast, very fast. And dead silent. The IS is quite effective. I would buy it separately.

ISO3200
f5.6 (wide open on the stm at this focal distance)
1/40 shutter speed
No noise reduction needed. Just processed the raw file in Canon's dpp.




ISO1600
f5.6 (wide open)
1/80 shutter speed
No noise reduction needed. Just processed the raw in dpp.





Low light video focus test. I can't imagine it focusing any better or quicker. No hunting. Just snaps right on target and holds it. I have it set to focus only on the center point, not the entire frame. I intentionally move around to make it have to refocus. At 30ish seconds in she moves into a shadow that's almost completely dark. It still tracked perfect. I don't see myself needing a dedicated video camera anytime soon.

The settings should tell you about the lighting.
Manual mode
ISO3200 on the first two clips. ISO1600 on the last.
f5.6
1/60 shutter speed

Do not watch if you are squeamish! It is a snake eating.



Same vid on Youtube

http://youtu.be/3kOCq6esn7o

Matthew Saville

Paul Iddon wrote: »

I don't see how any company would release a model in time for a specific sports genre. There is no NFL in the UK, we have soccer as our national sport and I'm sure Canon UK wouldn't guide Canon to release a model in time for the new season or the Champions League starting up in August/September. Release dates by camera manufacturers is when they want them to be, not to coincide with an American sporting event. Unless you know for sure otherwise of course?

Well done on getting published btw. That's always a great thing to have in your portfolio.

Paul.

If you think the NFL / Superbowl is not big enough to time a camera release on an international scale, well, I dunno what to say without sounding like an annoying American. Suffice it to say, the US is a pretty well-populated country, with a serious addiction to consumerism. Even if the corporate goal is only to sell beginner DSLRs, the fanfare of releasing a flagship in time for any fall sports season, or any sort of highly televised championship, makes this a very likely scenario actually. If not the NFL in particular, at least the combination of the NFL plus other sports which seem to all get underway around the same time in late summer / fall...

=Matt=

Matthew Saville

In other news, I had the good pleasure of testing the 70D's RAW performance the other day with the Sigma 18-35 1.8, and suffice it to say that image quality does indeed deteriorate roughly the same as all other Canon 1.6x cameras between 3200 and 12800, with maybe a marginal improvement. But certainly no leap or bound has been made. Test results coming soon.

=Matt=