New test scene beta begins with Panasonic Lumix DMC-GX7 samples


We’ve been working on a brand-new studio comparison scene for some time, and we’ve decided to give you a sneak peek, as a beta, using images from the new Panasonic Lumix DMC-GX7. We’ll be speaking more about this new scene (and the new interface that goes with it) very soon, but for now we wanted to let you have a look, try it, play around and tell us what you think.

We will, of course, be adding an increasingly large selection of existing models to the widget in the coming weeks, so that you can compare to older cameras, but we wanted to be able to show the GX7 results and invite feedback as early as possible.

This is a beta, so we’re aware there may be some bugs at first (including rather slow loading, first time it’s used). Please bear this in mind when responding. The best way of reporting an issue is via the feedback page.

New Features:

The two most obvious features should be the buttons at the top of each widget:


These buttons allow you to swap between a daylight simulation scene – which is the way our current scene is lit. This gives a good idea of how the camera will behave in good light. However, to give a clearer idea of how a camera might behave in the less-than-perfect conditions encountered in the real world, there is also low-light mode, partially illuminated using a tungsten bulb.

Image Size:

Another feature, aimed at making it easier to compare camera with different pixel counts, is ‘Image Size’ feature, that normalizes images down to an ‘Print’ size (7 or 8MP), and ‘Web’ size (~3MP), which puts all the cameras on a similar footing.

Panasonic Lumix DMC-GX7



Thanks for the effort and thought put into upgrading what dpr does so well.

No comparison can ever be perfec, I’m sure. But this is better than perfect, it is useful.


I like this new scene better, lots of areas with more details you can compare with. Hope DPR will find time to do tests again to add all cameras currently in old lab scene with this new scene.


Dear DPR Team (Mr Britton, or Mr Butler, if you’re around I have a question below…),

Print mode is a great addition.

I do have a question though, and I did not go through the 5 comment pages I admit, so apologies if it has already been answered:
How does it work?
My display has a certain pixel density, hence in photoshop, I have been obliged to scale the print mode function of that vertical and horizontal pixel density.
So is it actual print size when I see it, or is it print size on a 72 or 90 or 144 pixel/inch display equivalent?

Then I second any comment about a piece of fabric or fine repetitive texture, it’s good for judging apparent resolution and aliasing.


Mike Sandman

The widget is a great addition. The ability to compare cameras is very helpful. I agree with one of the earlier comments asking for a bit of fabric or other material with some texture. And now I know what to do with my old Schilling notes


Comparing samples at 100% is pointless, so I’m glad to see the “print” option.
The RX100II looks good except for lens softness in the corners.


I am not sure whether this new studio scene is so good. For example those “grass” areas are placed too far away from the centre, so any smudging might be attributed to the poor lens performance in the corners rather than as a result of aggressive NR.


Close call, All 3, except RX100II (for the sensor size) are good. But there is now no real advantage of APS-C compared to M43. Oly EP5 still looks warmer and slightly sharper of all. But that’s down to processing. Anything more, we are just ‘nitpicking’ here. All these camera’s have come a long way even compared to the D200, D300, 40D & 50D (that I still have). Did they not take good Pics? I am not comparing to Full Frame at all.


That little patch of greenery is the most telling, as it seems to confuse the snot out of the NR routines. The Sony NEX shows artifacting, The GX7 shows smearing, the EP-5 handles it well with just some resolution loss, and the RX100 II smears it out as well, but not as bad as the GX7. Of course this is all at ISO 12800, a setting I have never used.

Still considering the sensor size and resolution… if I have to look this close to see a flaw consider me really impressed.

Of other Note: Greenery is cyan-ish on Sony, yellow-ish on the m4/3 which is consistent with my experiences with the same WB settings.


Pretty impress with the GX7 it’s clearly showing nex 6 has no advantage even though its equipped with a larger sensor and EP5 don’t look so hot against it. Even the RX100II is impressive. Would be the second panasonic camera I really like to own after the L1.

Just Having Fun

The GX7 also has IBIS that works with every lens (but not for video) unlike all the NEX cameras. Since the JPEGs are virtually equal you have to give the edge to the GX7 overall.


Funny when you were trolling in favor of D600 several months ago, you were constantly telling people how IBIS doesn’t help when you need faster shutter speed to stop motion blur.

Tonio Loewald

I’m not thrilled by the use of photographs (are they analog photos? Dye sub prints?) The use of difference races and sexes is good in principle, but they all have dark hair. In any event these are photos of photos, so it’s not enormously informative.


Canadian coins on an Austrian 20 schilling note. Certainly has an international flair! The penny has been withdrawn from circulation in Canada, by the way…


Excelent work DPReview!
I am very happy to see the RX-100 II slightly better than the m4/3s up to ISO 400! The best pocketable camera ever!


Look at the RAW files – I don’t think that’s true, impressive as it is for such a small camera.


“Look at the RAW files – I don’t think that’s true”

Put the kit lenses on all three larger sensor cameras and reshoot this scene, RX100 will probably beat all three.


I would prefer more stuff prone to moiré and more red.


As far is I can see GX7 is slightly better than Oly P5 and Sony Nex-6 in both raw and jpeg (I checked 200, 1600 and 3200 ISO).
Canon 5dIII is much better from ISO 1600 up. Very impressive Panasonic. If the price drops 30% or more I’m interested. I would first like to see the EVF in action, though.

Btw: does anyone know which lenses were used?

I like the new studio scene, but I must say I miss the globe that was on the old one. That item was the best piece to compare differences.


I don’t see all that much difference between the Oly and Pan [not what I was hoping for as I’m probably going to be buying the GX7].

The 5DIII beats them both by a good margin – but you’d be a fool not to expect that. What surprised me was that the Sony NEX didn’t look any better even at low ISOs and probably a bit worse at higher sensitivities. All in all quite impressive performance from the new camera and the real-world high ISO samples actually look pretty useful for some things.

Marc Heijligers

I do miss some ingredients in the new test setup, that I appreciated so much in the old one:
– The small world globe (showing plastic texture and resolution)
– The batteries (same + contrast and color renderation)
– The pink face with small details (for detail rendering)
– The fluffy balls in the box (for low-ISO performance)
– The logo at the top of the bottle of the Martini bottle (details)

My recommendation is to remove some of the objects that are double in the symmetric setup, and to replace them with some of the old objects.

What I’m missing in general are
– Objects with texture, like paper / metal foil / glass / fluid / wood / stone to get a better idea of micro-contrast handling.
– I always like very much the red napkin with flower print shown at imaging-resource. Most objects that you use are high contrast, now showing these subtle differences easily.

Last but not least, a diagonal slider to see where exactly the focus / DoF is situated.

Steen Bay

Things from the old studio scene, like e.g. the small globe, would be much smaller now, because the new scene (and the shooting distance) is much larger.


Oh, DPR, you switched from Zuiko 50mm F2 to M.Zuiko 45mm F1.8. Care to comment about the change? That might interest many users.


In the past Dpreview used the 50mm f/2.0 lens to test every camera with a 4/3 sensor. And they even continued this with M4/3 at first.

But doesn’t it seem logical to use a native lens rather than one with an adapter?

Dpreview once called the 50mm f/2.0 lens a “must have lens for every 4/3 user.” The 45mm f/1.8 has become the must have lens for every M4/3 user today.


NEX-6 shots are with 50mm with an adapter. Do not see anything wrong with that. After all, it is a benchmark for the camera, not the lens.

Though, yes, use of a native lens make some sense too. Just like the print and web sizes/different lighting benchmark, gives an idea of the realistic IQ. But then it should be realistic for all other cameras too.


I appreciate DPR effort on the new studio test scene. And there are things to like about it. (Though I personally prefer the very busy overloaded old scene. But I can imagine how hard it could be to maintain it over years.)

My concern is that, as usually, it is unlikely that you would reshoot older cameras with the new test scene, making it impossible for people to compare their owned old gear to new one. That makes it hard to see the IQ progression.

Also, many (esp mirrorless) cameras were never given a shot (pun intended) at the old studio scene. Panny sensors made small steady improvements over the years – but that is absolutely not reflected on DPR. I think if you change the test scene now, probably you should at least consider looking for ways to make the scene shooting process faster (or two shooting processes? – full vs. preview modes?), allowing you to publish the image data even for cameras you do not deem important enough.

rare wolf

sorry, double post

rare wolf

Where are the specular highlights that were in the previous test image (paperclips)?

rare wolf

Where are the specular highlight/detail that were in the paperclips?


Regarding the scene:
Having twice the brushes, color tubes and cards in the scene doesn’t make a lot of sense to me. Instead I would prefer additional objects, something like the globe (of the old scene). The different text sizes on the globe were nice for comparison of resolution and high ISO capabilities of different cameras side by side and which texts can still be read. Different text sizes are relevant to allow also a distinction or comparison for the future when the cameras in maybe 4 years come with the a 100MP sensor.


If you haven’t noticed there is plenty of text in these scene. Far more than the older version


Two points–first, there is more text of various sizes in this scene than the old one. Second, the repeated elements are very useful because they permit comparison of bright and shade performance, which is a major improvement over the old test scene. (check the tungsten version.)


“Having twice the brushes, color tubes and cards in the scene doesn’t make a lot of sense to me.”

Check the “Sun”/”Lightbulb” lighting buttons at the top of the comparator.

Does it make sense now? Pretty nifty addition.


I think you should have included elements from the previous test scene for direct comparison with the new one, like you did last time you revamped it. Now it will be so much harder to compare older cameras with new ones until a few years have passed.

The second option is of course to shoot the new test scene with all the older cameras still available on the current market, even if that won’t include discontinued models that people still might have liked to see compared with new ones.


Sharpness seems to be slightly more fuzzy in Daylight than artificial light (looking the bottle area).

And maybe for pixel peepers a coordinate bar (like in landmaps) to communicate issues one sees in photo.

A bit sad that older cameras will now become uncomparable. (good news for manufacters since image quality improvement seems to stagnate last 2 years )

And one request I would like to see; A test photo with 1 sec exposure bulb light. (long exposure noise at every iso).

But after all the point the test image seems worthwhile the effort to trash older test data. The image makes sense and points of IQ attention are easy seen. (older version was more or less a forest where each and everyone had its own favourite point to inspect)


The lightning of the scenes is very different for the different camera models.
At this moment it’s not yet good to compare.
Hopefully all scenes will be shot one more time, using the same setup.

Steen Bay

Looks good, nice work. The only remaining problem now is that the exposure (the amount of light/photons hitting the sensor) can vary quite a bit between the tested cameras. For example, according to DPR’s ISO accuracy tests, then the ‘measured ISO’ on e.g. the E-M5 is 1/3 stop lower than indicated, and on e.g. the NEX-6 the measured ISO is 1/3 stop off in the opposite direction (1/3 higher than indicated). That means that if comparing the two cameras at e.g. ISO 1600 with the OOC JPEG brightness matched (like DPR does), then the exposure will be app. 2/3 stop higher for the E-M5 than it is for NEX-6, and that’s of course not quite fair (because the exposure determines the amount of shot/photon noise in the image). The problem could (theoretically) be solved by matching the exposure (the lighting, shutterspeed and T-stop) instead of the OOC JPEG brightness, and then adjust the brightness of the RAW/JPEG images afterwards.


there is no problem that E-M5 got multiple tricks to cheat users. the very idea behind 4/3″ is cheating from beginning.

we should not rely on camera advertised ISO at all, but accurate exposure on the sensor which is difficult but we should try our best.


1/3 stop is within specs, and so tiny that I won’t lose sleep over it.


“The problem could (theoretically) be solved by matching the exposure (the lighting, shutterspeed and T-stop) instead of the OOC JPEG brightness, and then adjust the brightness of the RAW/JPEG images afterwards.”

Yeah, and then you have to first measure the T-stop of lenses and then tons of work in post, just for 1/3 stop difference?

No thanks

Steen Bay

P.S. – If it isn’t possible to match the exposure (or if it is too much bother), then the next best thing would be if it was easy for the viewer to see what DPR’s ‘Measured ISO’ is for the tested cameras, like e.g. ISO 160 for the E-M5 and ISO 250 for NEX-6, at ISO 200. It would also be possible to calculate the actual difference in exposure by comparing DxO’s ‘Measured ISO’ (which is a different thing than DPR’s ‘Measured ISO’) to the saturation of the RAW files.

Steen Bay

@ET2 – The differences add up, and in some cases the difference in exposure can be more than 2/3 stop, which I’d consider to be rather significant.

Andy Crowe

When they add the display of shutter speed and aperture to the results (it’s planned apparently) that will show if there is any large difference in iso/metering.

Steen Bay

@Andy Crowe – Yes, that would help, but the f-stop (and shutterspeed) will probably vary, depending on the format (which makes it more difficult to compare quickly), and will the lighting be constant? It’ll be much faster/easier to compare DPR’s ‘Measured ISO’ (or even better, a ‘calculated ISO’, derived like described above).


Dear DPR

Congrats on the new scene. You guys sure are working your butts off lately, and it doesn’t go unappreciated.

My suggestion for today, is related to the time-intensity and man-hours required for such systemic updates as the new test scene.

As you would be aware, in many other popular industries, especially sports, the use of volunteers is widespread. Have you considered theuseof volunteers, even on a short-term or project basis, to exponentially multiply your resources and “bring to market” new features, tests, and (gasp lol) maybe even reviews in a much quicker timeframe?

Surely the massive pool of forum members provides you with a (free) and invaluable resource, that is sitting virtually untapped on your doorstep? The varied experience and skillsets of many members should not be underestimated.

Just a thought.

Greg 🙂


Yes, congratulations to DPR. But do you think your tought could be trustworthy? I doubt.


Does it address dynamic range?

Stu 5

Not by the look of it and it would have been simple to do.

Martin Datzinger

I welcome that you redistributed everything to get it more in one plane and generally smaller because we thankfully don’t live in a 4MP world anymore. Just wondering how you’ll deal with backwards compatibility. Reshoot with every still sold camera? Seems like a lot of work to do!

I’ll miss the colorful feathers of the last version, I can live with the new color resolution test patches though. What bothers me is that there seems to be no shadow performance test such as those those overshadowed puffy little balls anymore. The faces? I don’t know what to judge from them. They are obviously prints, which means DR limited, off-colour and thus very unnatural looking in the first place. I guess the symmetry is to rule out lens misalignments? Please use the ColorChecker Digital SG instead of the (unnecessarily big) simple version, that is contained within the SG anyay. One last thing: Could you make the whole thing retina screen ready? Pretty please?

Andy Crowe

If you switch to “low light simulation” view then the bottles seem to make good shadow performance targets to me.


I guess the point of the faces is to see how the cameras handle skin tones. How do you know that the prints are off-colour and unnatural-looking? I seriously doubt that DPR would have used them if they were. Have you never seen skin tones faithfully reproduced in print?


Not directly related to the new scene, but I do think it is very important to come up with a test for shutter shock/ mirror slap. After all, what is the point of spending £1000s on a new system only to find that a lot of the shutter speeds are unusable. What is the value of hi-res lenses and good high iso capability if many shots are blurred due to SS or MS?

I know SS may be lens dependent and the tests will be difficult to devise, but you are the review leaders and it would be a real coup to lead the way on this thorny issue!


Apologies re my comment about not globally selecting RAW/JPGs & ISOs. The new system is even better than the old!


Τοο empty I say…
Meaning that u could have all these in less area…
Quite functional though and more comprehensive…


Please add a Nikon D600 as a Nikon benchmark.


D800 I would prefer. actually I would like to see a stitched one to know what are really there (like 4 x 4 D800).


Add Martini bottle from the old scene please.. as by standard, image comparison always open on that medallion/coin image of that martini bottle


Its good to see new developments and, especially, the opportunity for feedback – much appreciated!

Perhaps I didn’t spot it but I always found the facility to ‘globally’ change the ISO and RAW/JPG settings very useful. Its a bit of a drag to have to change each one individually. Apologies if I’ve missed it!

Secondly, I do miss the old scene. I’ve used it many, many times and I feel I know what to look for and how to interpret the results.So could you include some of the original items which I find are particularly useful – e.g. the watch, the note with the old lady’s face, the bottle label with the coins etc.

I know I’ll get used to the new scene eventually, but could you explain why it was necessary to use a completely different one and what the advantages are please?

Many thanks!


Matt Photo

(duplicate in error)


If that would be possible, it would be good to know the distance from the scene that a photo is taken (as 28mm fixed lens cameras
-Sony RX1,Fujfilm X100S or others – get a lot closer to the scene gaining more detail) OR prefix positions for 28mm, 32mm (for fixed camera-lens) and 85mm like you do.
What are the sizes of the chart ?

Andy Crowe

IIRC the scene is framed to capture its full height so there’s no detail benefit for being closer to it, tho m4/3 benefits slightly in size for being 4:3 rather than 3:2.


we will need time for DPReview to realize the “fit to picture height” problem. they have been doing the wrong way for long time.

for the different angle of view, they chose a “flatter” target to minimize the problem.


Using different lenses in order to fill the frame, you use different distances from the chart, specially when 28mm lenses is used, where the camera is extremely close therefore you benefit in detail.


basically best tele should be used as much as possible.
I may choose Canon 100/2.8, Nikon 105/2.8, ZD50/2, …
for targets that have depth.

we usually shoot at 20-40 times of focal length for lens test.

Andy Crowe

@Petrogel detail is based entirely on lens sharpness, because the actual parts of the test chart will be equally sized in terms of what the sensor is capturing. In fact longer lenses are generally sharper than wideangle lenses.


No Andy !! detail is a combination of lens-sensor-software coordination. Being closer to the object gives a lot more benefits specially for cheep wider angel of view lenses

Matt Photo

Why do you display the image at only 565×425 (or whatever it is) pixels?

It leaves most of my screen wasted – black – not to mention my second monitor.

Surely in this day and age we should be striving for adaptive rendering to make as much use of display area? On my monitor, for example, it could be more than twice as big in each dimension and still leave space for the comparison widgets on the right-hand side. (On Connect you could make it crappily small for smart phones etc 🙂 )

If you are doing a new tool why not make it take advantage of the high-DPI devices which are coming on the market already?


there might be a simple flaw in the test
if DPReview had fit the target to image height.
this wastes 12.5% of area on a 3:2 sensor (blank to the left and right),
which will be 0.17 stops under evaluated than a 4:3 one.

I commented on the issue before.
the right way to do is “same portion of area” that
if one subject occupies 10% of area on one sensor,
it should also occupy 10% on another, regardless of aspect ratio.

if you would want to do it right with reference to picture height,
simply fit 1 unit length to 3:2 while 1.06066 to 4:3 (sqrt(9/8)).


picture height is a handy tool for analog TVs but it was not designed for comparison across different aspect ratios.


> this wastes 12.5% of area
sorry should be 1/9 = 11.1%

Andy Crowe

There will be no benefit to noise at 1:1 viewing though, the only difference will be that details are that little bit bigger and thus a tiny bit clearer.


@Andy Crowe, please think it again.

Andy Crowe

@yabokkie how does a different framing affect the amount of noise at 1:1 viewing then?


I said think about it again and you didn’t.

Andy Crowe

Think about this, put the preview patch over one of the blank part of the test and crank the iso up. How does the fact that the scene is framed vertically affect the amount of noise you’ll see?


different area under the same lighting condition,
and you say no effect on noise, what’s wrong there?

DPReview should have noticed it that their tool doesn’t compare the whole frame directly whatever the aspect ratio or even the target is misaligned (a lens issue maybe). we only care and compare same portion of area.

we know the result of whole frame if we know the result on a same portion of area in the frame.

Andy Crowe

yabokkie the cameras are metered to expose the scene equally you silly billy. I’ve checked in Photoshop and the brightness on the various patches of grey are as near identical as possible. If anything the NEX6 is metered slightly brighter which is actually giving the APS Sony a slight noise advantage.

It seems that you’re so obsessed with trying to prove everything through physical measurements that you haven’t even looked at the images to make sure what you’re saying is true.


you are saying that sensor area has no effect on image noise at the same exposure and you insist you are right? (DPReview “cut off” 1/9 of a 3:2 sensor before comparing it against a 4:3 one)

btw, do you know that image noise is measured against (a portion of) sensor/frame area? remember we are talking about image quality not pixel quality.

sorry I might be wrong by saying “think” to you.

Andy Crowe

Sensor area obviously has no impact on noise levels at 1:1 pixel level. If you had a 12mp m4/3 sensor and a 48mp FF sensor built with the same sensor tech then at 1:1 viewing the noise levels would look identical. Obviously the scene as a whole will have less noise and more detail on the FF sensor because there are more pixels, but the actual noise when viewed at 1:1 will be the same.

The DPR test scene viewer happens to use 1:1 at its highest zoom level, hence m4/3 cameras will get a slight apparent resolution advantage but no flat-colour patch noise advantage.


so you are talking about pixel quality.
then, how you convert pixel quality to image quality?
or you are not interested in image quality at all?

Andy Crowe

I was talking about the effect of the different sensor ratios on the studio comparison viewer, which is quite different to the effect you were describing (i.e.there is no noise advantage but a detail advantage instead). What you said is actually misleading to people trying to compare the GX7 and APS cameras using this studio comparison tool. Obviously comparing images at 1:1 pixel size can be misleading, especially when dealing with cameras of vastly different resolutions (like the NEX7), but it’s better to discuss what this means for the viewer at 1:1 pixel viewing rather than taking about the effect on equally sized images, which the studio comparison tool doesn’t show.


so you have no interest in image quality and
I have no problem with that.

I’m interested in image quality.
I’d prefer pixel quality as low as possible (smaller pixels).

Andy Crowe

I’m just interested in discussing what the studio comparison viewer actually shows and how that relates to image quality, not just calculating image quality based on sensor size and distance to the test scene.


okay you are interested in image quality which is great. then may I ask again how you would know image quality based on your study on pixel quality?

… and what do you mean by image quality?

Andy Crowe

In fact I’ve noticed that if you select to view the scene at print resolution then it resizes the image so all elements on the page are identically sized, so there you go they’re giving us both the 1:1 pixel comparison and a like for like overall image noise/quality comparison, no need to fuss over the exact framing of the chart.


Well done. One suggestion: I would be pleased to see one detail of the old studio scene again in the new one – to be able to compare the “old” cameras and the new/future cameras. F. E. the face of the lady in the red bank note.


It is quite noticeable the variation in image quality (noise and sharpness for example) in different areas of the scene. Sometimes, the LX7 is clearly superior to the others. in other parts, the Olympus, and in still others the Sonys.

This is why one cannot make a statement that Camera ‘A’ is best. It should be qualified as best under these conditions and then list the conditions.

One thing that I’d like to suggest is to offer some form of noise reduction (at least colour noise) in the RAW previews at high ISOs. We all know the images can be greatly improved. As it is, to get rid of the noise distractions, we have to download the files and normalize them ourselves in order to make a fair comparison. Thanks.

Andrew Butterfield

That’s a very good point. The RAW images in the tests are not what the user would expect or get from their own workflow. It does give a useful idea of the base level of noise, but having a standard amount of noise reduction for comparison would also be useful.


some users don’t care about image quality and use JPEG,
some users care a lot and use RAW.
DPReview’s tool might be designed for those who don’t care.

Andy Crowe

yabokkie they’re giving us a tool that can show raw and jpeg at both 1:1 and like-for-like sized, what more do you want?


Interesting (and perhaps not surprising, considering there is no AA filter) to see significantly more moire with the GX7 compared to EP5. Just look at middle row of coloured circles next to the bottle of honey vinegar (6 O’clock position).

I somehow doubt GX7 would be great for macro photography. (Seeing these results, I personally wouldn’t mind seeing an AA filter on this sensor). It just shows you, with the size of this sensor, you really need the AA filter (this is just an amateur’s observation – please take it that way).
Nevertheless the GX7 withs it’s form, design and tilting viewfinder it’s quite an appealing camera. Although not planning to buy it, but well done Panasonic.


wait – there’s no AA filter on the GX7? the Oly is showing better sharpness and contrast. I thought it may have been because of the aa filter… but there is none.

so Panasonic really doesn’t have an answer to the Oly sensor.
or is it just a bad lens causing these results?


> so Panasonic really doesn’t have an answer to the Oly sensor.

so Panasonic really doesn’t have an answer to a sensor made by themselves?

Andy Crowe

If you look at the text samples (12 o’clock just above the central grey square) the GX7 text looks slightly sharper and has slightly stronger moire than the Olympus. (Make sure you compare RAW btw, Panasonic and Olympus’ JPEG moire removal seems to be more effective than ACR’s)


Why no shutter speed info? I think that it’s a key info. How can I judge a camera in low light with same iso setting but with double shutter speed for example? Now I have to download the images and read exif, if you add this important info, than will be a lot easier compare different camera’s.

R Butler

We will be including the shutter speed, aperture and possibly illumination level of each shot – that feature isn’t complete yet.

victor china

that would be very handy!


definitely no test should be based on ISO.
would appreciate if you calibrate shutter speeds, too.

Andy Crowe

yabokkie are you really so obsessed with measurements that you’ll only accept measured and calibrated shutter speeds? Do you calibrate all your own cameras’ shutters too?


I used to do it quite often when I had analog TVs at home (some of the last models won’t work well though).

the test result is not reliable if the testers don’t have good control of conditions. isn’t that simple enough?

Andy Crowe

You’re not going to get a very scientific measurement using a CRT TV, and why go to the trouble of spending all that time measuring the minute tolerances of shutter speed when the transmission loss from the lens will be a great deal bigger? What other review sites have ever measured actual shutter speed?


can you please add stuff from the old scene? i really miss the watch, the pins and mickey. i always found them very helpful.

Barney Britton

The scale would be totally different (thus of very little help). This new scene is many times larger than the old one.


@Barney: is it available for download, at least, so we could test the cameras that we have, to compare them with the new ones you test?


Add Martini bottle from the old scene please.. as by standard, image comparison always open on that medallion/coin image of that martini bottle.


I always looked at the salt shaker, pepper, cloths, and the gauge dial.
I’m going to miss the old one.


re: GX7 vs EP5

is the poor sharpness and lower contrast on the GX7 due to an over-aggressive anti-aliasing filter that covers it’s sensor or due to the lens used for the test?

what lenses were used for each of these cameras?

I’ve seen youtube videos where they try to establish whether it’s more important to have a good sensor or a good lens, and the lens comes out on top. so these tests don’t really tell us whether the difference is due to sensor or lens. dynamic range, colors, color channel, noise, could be evaluated with these, but sharpness and contrast are very lens dependent.

maybe these tests should start to be done with different lenses: a kit lens, plus a good prime in the 35mm/50mm equivalent range. if a maker’s kit lens is notoriously below average, it wouldn’t show the camera in it’s best light… and we know that is the case for some kit lenses, and these tests mislead us to believing we are comparing cameras and not lenses.

Oleg Vinokurov

Maybe it’s the jpg engine? I see the opposite, gx7 is a bit sharper. But lens is same. GX7 has no AA filter, this could be the reason why GX7 looks a bit sharper in raws


the GX7 has it’s ISO at 125 – whereas the EP5’s is at 200 by default. I brought it down to 100, and everywhere I clicked the Oly looked better (sharpness and contrast).
go check out the sponge and feathers close to the corners (with both at iso100).


it’s very careless to compare cameras on ISOs.

Oleg Vinokurov

You sure changed to raw? Even when comparing extended iso 125 and 100 on gx7 and ep5, still gx7 a bit ahead in terms of sharpness and contrast. Check out hair for example, really fine detail. Maybe mixing up gx7 and ep5 samples?
In green feather e-p5 is soft and blurry, where gx7 still retains some detail.


@DPR: I’m glad to see that you shot these with some other cameras for reference. however, you should add a few more in there, especially a wider selection of the higher-resolution cameras from Sony, Nikon, Samsung, Pentax, to keep the readership here happy… the best of each maker for reference in both APS and FF formats.

Barney Britton

obviously… we’re working on it 😉 but it takes a long time.


I think either do what they are doing now, GX7+EP5+5D3+NEX6
or publish the tool 1 month later with more cameras

which one do you prefer?
I prefer the current approach, at least it helps people/me to pick GX7

Scott Birch

Some of us are aware of the scope of your effort. Thank you Barney. That said, may we have the moon on a stick?


For E-P5 in the comparison, “Keep Warm Color” was on, wasn’t it? 😉

David 247

That was answered elsewhere by one of the dpreview staff, but I believe the answer was that it was whatever the camera’s default is, and I think for the E-P5 the default is “keep warm color”. Not sure though.

R Butler

David is right – the E-P5 is set to ‘Keep Warm Colors,’ because it’s the default setting.


Finally! Finally you use AWB for low light test scene, just like most people will use it. Now, next step please – KIT LENS, just like most people will use it! Otherwise, we will never see better kit lenses! And moving scene. 🙂


I have to disagree about kit lenses for a test shot. Give the camera the best chance with the best lens/aperture because that’s what pixel-peeping photographers will do. Kit lens fans won’t care nearly as much!

R Butler

We may include a set of shots taken with kit lenses to show how much variance you can expect from this ‘ideal’ situation. We’ve got a lot of cameras to shoot first, though.


The desk top image is fine, but we need more. dpreview should take full advantage of the person on their staff they hired who had the internet camera test site For each camera he tested he posted identical landscapes taken from the same spots in San Francisco and Palo alto. I have saved hundreds of his real world test images.

Lets have the Space Needle scene that dpreview has used from time to time (taken from near the Cancer Center building in Seattle) be one of the standard scenes. The present system of comparing cameras to each other by using shots of a baseball game for one camera and of a leaf laying on the grass for another is not very useful for landscape photographers.

Another site that takes the same landscape for each camera is

Dpreview did this several years ago for their review of compact zoom cameras. Each had a landscape of the Space Needle in the distance with lots of other Seattle buildings.


the very significant problem with that, is that the sun is at a different height at the same hour at different times of the year. it is IMPOSSIBLE to obtain consistency and equal conditions to hare reliable comparison shots. it’s a valiant effort, but oh so flawed. pretty useless when you think of it.
but I do agree with your point, that you can’t compare cameras with different scenes/subjects… however it doesn’t work with the same scene either.
that is why they do the studio shots. it’s not perfect, but at least it allows you to control the conditions and make sure they are the same, so you can compare.


They absolutely should have bigger test scene, at least human-sized, with shooting distance and FLs to match.

Simon Joinson

it is pretty big (like 6 feet wide)

Jeff Keller

Hey, I like that idea! Too bad the weather in Seattle is much less consistent than in SF (and that’s saying a lot).


The RX100 II is spanking some serious butt.

In raw of course.

Shangri La

I hate judging image quality based on pictures of pictures of unknown quality.

Source Article from