D40 versus D3000 - Revisited

<p>I posted some comparison results in Anike's "<em>Which Camera to Buy? D40 vs. D3000 vs. D5000 vs. D90</em>" post yesterday. While I feel the out-of-camera files clearly show the D40 to give better image details and should be her camera choice, I think my discoveries in stripping NEFs in Adobe Camera Raw moved outside the range of her question.</p>

<p>But I would still like to discuss this a bit with The Regulars. Ultimately, this is a follow-up to my "D40 verus D3000" post on September 30th..... <a href="../nikon-camera-forum/00UcfD">http://www.photo.net/nikon-camera-forum/00UcfD</a></p>

<p>I have compared files from the D40 and D3000, shot under similar conditions, in a dozen different ways. For the most part I tried to keep the camera settings and manipulation as low as possible. Whether ISO 200 or ISO 1600, the D40 is the clear winner. Less noise (even at ISO 200), more detail, better prints. In response to a request in Anike's thread yesterday, I posted these two 100% crops from ISO 1600 images from both cameras.</p>

<p>This is the D3000's JPEG crop:</p><div></div>

<p >And this is the D40’s JPEG crop:</p><div></div>

<p >Since both cameras had overdone the colors a bit (the tile boxes are more maroon than red), and the WB was slightly different between the two, I decided to feed them through Adobe Camera Raw and equalize the WB.</p>

<p > </p>

<p >When I had them in ACR, I also tried stripping all the JPEG-making tags from both of the files. The result was enlightening. BOTH camera files improved by a quantum leap (or three). It was plainly obvious that even with my very basic camera settings, the cameras were making some serious changes between sensor and JPEG file. ViewNX did the same thing, as one would expect, if asked to process the NEF into a JPEG.</p>

<p > <br>

This is the data in a similar crop from a JPEG made in ACR by setting all sliders to zero. First the D3000:</p><div></div>

<p >And then the D40’s “stripped-in-ACR” JPEG:</p><div></div>

<p>As you can see, the data in BOTH is much, much better. Compared to an in-camera or in-ViewNX JPEG (like the first two ‘camera JPEGs’ in this thread), it looks like the pictures come from different cameras. The D40 still holds a slight advantage, but the gap is much narrower.</p>

<p>Am I missing something basic and huge regarding setting the D40 (and D3000)? My D40 settings were zero’d Tone, Saturation, and Hue, with –1 (on a –2 to +2 scale) Sharpening. My D3000 settings were zero’d Contrast, Saturation, and Hue, with ‘2’ sharpening (on a 0-9 scale). Changing sharpening to minimums had little improvement. According to ACR, there is quite a bit of Brightness and Contrast dialed into the Nikon JPEG mix, with a significant Tone value, too.<br /><br />Does anyone know if there is any way to get a truly neutral JPEG out of the D40 (and D60 and D3000….)? The “lack of data” is there, as the ACR-stripped NEF file shows, but I see no way to get that out-of-camera. Even at it’s lowest settings, the effect of the camera’s JPEG processing is HUGE. Why would Nikon discard such clean file information in both cameras??</p>

<blockquote>

<p>Even at it’s lowest settings, the effect of the camera’s JPEG processing is HUGE. Why would Nikon discard such clean file information in both cameras??</p>

</blockquote>

<p><br /> These cameras are targeted toward consumers and entry level photographers who believe they want a dSLR for access to interchangeable lenses rather than a compact P&S type camera with a non-interchangeable zoom.</p>

<p>Nikon, like Canon and most manufacturers, understands this market. Most buyers of these cameras want photos that will look good straight from the camera. They want JPEGs that don't need editing to look good online, or that can be tweaked as desired at those do-it-yourself kiosk front ends at minilabs for prints.</p>

<p>Most of the criticism of the image quality of entry level dSLRs comes from folks who'd be better off with higher end cameras. They're usually criticizing a camera for doing what it's actually intended to do. Sorta like criticizing an apartment sized chihuahua for being tiny and not big enough to defend the ranch against bears.</p>

<p>But the Chihuahua should still be able to bark and alert the ranchers.</p>

<p>Thanks, Lex. I am aware of what you speak of, and it's the main reason the Canon dRebel ate the D70, and why the D40, in turn, ate the dRebel. You'd just think 'default' would provide the out-of-camera, saturated pics for most buyers, while menu choices would allow dialing it out (or up, if John Q wanted that).</p>

<p>well put, lex.</p>

<p>that's why some of us have different dslr for different occasions. when i visit my granddaughters to play with them in the park, etc., or just have quality time indoors or in the backyard; or in the pool, i use my d60 (used to be the d40 but my eldest granddaughter has it now). this way i can just pull out the card and give it to my granddaughter and the images get downloaded right away to her computer. she is good now with cropping and adjusting brightness and contrast. that's all she needs for straight-out-of-the-camera images from entry-level dslr.</p>

<p>of course on their baptisms (and similar more important events) and school sporting events i use a better dslr suited for the occasion. then when i go out on shoots where i get paid, there is another set of gears.</p>

<p>thanks for all your effort, joe a. but hey, i love my d70 (my first dslr). i'd say when the rebel tried eating the d70 it choked :-)</p>

<p>Joe,<br>

Thanks again for your sterling work.</p>

<p>I don't quite know what you mean by a neutral JPEG.</p>

<p>On the D40 I believe the default settings give a little too much saturation at the default AUTO setting and some of the reviews that use a colour checker confirm this (the D60 measures rather higher I recall and I haven't seen the D300 results yet). I'm not sure if it's any higher or lower than the default settings on other Nikons and would need to see measurements. I asked Nikon UK what the difference between AUTO saturation and normal was and they couldn't give me an answer that made sense. All the EXIF data I've looked at on mine says normal anyway.</p>

<p>Contrast is by default set to AUTO which means that the camera adjusts the tone curve based upon the subject and indeed the EXIF data I've looked shows a difference between flat subjects and contrasty ones. This makes it a little easier to avoid blown highlights with contrasty subjects and flat ones have a little more snap. The effect is quite subtle and I quite like it.</p>

<p>Sharpening is also set to AUTO by default and I believe I read somewhere that the intention is to reduce it at high ISO in order to reduce noise effects. I've not confirmed this by looking at the EXIF data however. In any event you thought that sharpening on the D40 was <em>higher</em> than on the D3000 which would tend to make the D40 noisier not less noisy.</p>

<p>In general I've been quite happy with the default settings for the photographs I take which are mostly landscape and I don't believe I'm a high saturation fanatic. I use JPEG almost exclusively and when I try to match the saturation and contrast of a JPEG with a raw converter I find that it is quite hard to match the good compromise between noise and sharpness. I only use GIMP and have tried several methods of noise reduction - some of these are supposed to quite state of the art but the interfaces are usually rather clunky and slow. I do know that without any noise reduction at all the results are significantly noisier than the default JPEGs.</p>

<p>This is why I find the apparent difference between your JPEG and RAW results rather baffling. I can only think that if you increased the saturation, contrast and sharpening to match that of the JPEGs then the noise might well be greater and the results more like those for the JPEGs.</p>

<p>I do know that to get saturation and colour fidelity up to realistic levels it is usually necessary to apply some matrixing as well as demosaicing and that this increases noise. I don't know if this has been done in your case. The matrixing is done to overcome lack of colour purity in the three filters - e.g. if the red filter has some green in it then you need to subtract some green from the red channel and this increases the noise level.</p>

<p>I guess I'm fortunate in that I'm not very interested in fiddling with RAW images and so my decision to keep the D40 is confirmed.</p>

<p>...And (heaven forbid) you seem to have confirmed more or less what KR said about the D3000 (!!!) I think that apart from his mania for high saturation the settings are fairly modest.</p>

<p>Yours rather puzzled,<br>

Richard</p>

 

<p>Oh - I should have said that I was using a Nikon D40 tone curve when doing the RAW conversion - without it the results looked very flat compared with the JPEG.</p>

 

<p>Thanks Richard.... By neutral JPEG I mean as close as possible to what the image stripped of all tags by Adobe Camera Raw looks like. A true zero file for the times I'd want it set there, and in the case of the D3000, the only real way I see to wring near-D40 performance out of the D3000. I have tried all combinations of settings, from 0 down to -2, as the D40 allows under Optimize Image > Custom. No matter what, there remains enough in the JPEG-drawing or compression algorithim to eat up the details the sensor actually sees at capture.</p>

<p>It IS easy to bring the JPEG back to the more intense (in a normal way) colors and image intensity of the camera JPEGs during ACR conversion, and the fine detail does NOT disappear as fast as the image improves. I certainly don't want to do this all the time to every image, but it's nice to know better data IS there, hiding, if I shoot NEFs. <em>Much better data</em>.</p>

<p>For me. both the camera-JPEG image quality and the ACR-stripped-JPEG image quality are better in the D40 than the D3000. The stop in Adobe Camera Raw closes the gap between the two, but the D40 still wins at the pixel level and the print level. The D40 stays, even though I covet thy CAM 1000 module.</p>

<p>Ramon.... I love the D70. Still recommend it and the D70s to people on a true budget that want two control wheels. I was referring to <em>sales</em>, not performance. Canon took the world with the dRebel, market-wise. And Nikon stole it back with the D40 (and the 55-200VR).</p>

<p>I just went to www.imaging-resource.com where they have saturation measurements for a variety of Nikon cameras. I looked at the saturation error for all (I think) of the models going back as far as the D40. These are all sRGB and I believe they are all at default settings. All the figures are percentages. Note <em>all</em> are oversaturated.<br /> D90 +3.8<br /> D5000 +5.3<br /> D3x +8.1<br /> D300 +8.9<br /> D700 +9.0<br /> D3 +10.3<br /> D40 +13.4<br /> D60 +18.0<br /> D40x +20.0<br /> So the conclusion must be that the D5000 is a <em>far</em> more professional camera than the D3 ;-).</p>

<p>Yes, the D40 stays. In my opinion, it still represents one of the best buys in the DSLR market when looking from a price/performance relationship. Only six megapixels but what a nice set of pixels. Producing neutral JPEGs is rather easy if you use Lightroom and shoot in NEF (RAW). It happens that Nikon has reserved some information about NEF files and Lightroom doesn't show the in-camera saturation or contrast adjustments made by the user. As a result, if you're a Lightroom user you always get neutral images.</p>

<p>But if you shoot in jpg format, then all the adjustments made in the camera tanslates directly to Lightroom because threre is no secrets hidden in jpg images to Adobe programmers.</p>

<p>lex, great analogy. as an aside, i was almost attacked by a vicious yet fortunately leashed chihuahua the other day. scared the bejeepers out of me until i realized the dog was the size of my two hands put together.</p>

I own both the D90 and D40. The D40 is slooow when chimping. I made the simple conclusion that the D40 uses a slow

embedded processor and RAW-to-JPEG hardware circuit. Nikon calls it Expeed, but it has to be one of the slower and cheaper versions. Being slow, it must discard more data than a faster processor in order to perform the JPEG conversion in real time. It seems that the D40 keeps more useful set of data than the D3000.

 

On Canon, the embedded processor used is the ARM, used in over 90% of mobile phones. It's so common that hackers made the CHDK operating system patch to upgrade camera features. Nikon doesn't provide such information so it's not possible to do a similar hack.

 

Just my view as an embedded systems designer.

<p>I find that when differences are small between cameras, they can easily be equalized through post processing, especially when shooting RAW. </p>

<p>Lex's comment "<em>Most buyers of these cameras want photos that will look good straight from the camera."</em> for this kind of camera is right on the money. I am taking an educated guess that if you printed 4 x 6 prints or even 8 x 10 prints from the unprocessed JPG files between the two cameras at high ISO, they would likely look identical or so close you would not be able to easily differentiate them. </p>

<p>I believe more advanced users who know the power of shooting RAW especially for higher ISO shots would not have any difficulty making higher ISO shots look great.</p>

<p> </p>

<p>Raden.... If you think the D40 is slow, you should see the D3000. Even shooting JPEGs, it locks up for 2-3 seconds after a single shot with NR on. Can't bring up the shooting display to take the next shot into the buffer. Can't open the menus. Can't play an earlier shot in preview. Nothing. The camera is dead. Does the same thing with NR off after a 3 or 4 shot burst.</p>

<p>Elliot.... The difference is not that small. With in-camera JPEGs, it's more like 1.5 stops. Only in the ACR-stripped file does the gap close, and I'd still give a 1/3- or 1/2-stop advantage to the D40. </p>

<p>There's no doubt in my mind that these cameras are designed to make 4x6 prints and I do not dispute that with you or Lex. When you consider how little megapixel resolution you really need for a 4x6 (what, 1 megapixel or so?), the JPEG algorithm in the D40 and D3000 has plenty of room to smash the sensor's data in favor of blobs of color. </p>

<p>But I don't buy Lex's blanket dismissal. Most D3 shooters are pros shooting ready-to-use-JPEGs; doesn't mean it comes without NEFs. Just because a product is aimed at 90% of the market doesn't mean it should ignore the other 10%. The D40/D3000 may be that way because of marketing, but that doesn't make it right. I still think the menus should adjust across a larger range. </p>

<p>Personally, I do not think it has so much to do with making Soccer Mom 4x6s out-of-camera. I think it has to do with not competing with the D80s and D90s in the lineup. The fact remains that the "little" 6MP sensor is a star, and I'd hope people would think twice about blindly recommending the D3000 (or a D40x/D60) to newbies simply because they are newer.</p>

<p>The downside is that the D40 is getting hard to get already (in Europe at least). So sooner or later, we will have to recommend the D3000 when budget is tight and people want new instead of second hand ;-)</p>

<p>For what it's worth, I do think Lex' remark on the "output tuning" of the consumer range is quite on the money. When the D50 was launched, it was one of the remarks how the default was more saturated and punchy compared to the D70, to match "normal market taste", so to speak. RAW converters do nothing different in their default camera profiles.<br>

Now with the Picture Styles on the newer cameras, the colour response and such are far more consistent throughout the range. But for low noise, high dynamic range, you end up using the neutral style, which frankly often enough looks a bit dull. It's all "give a little, take a little" in that respect.</p>

<p>Disappointing, still, this performance of the D3000. It looks worse than the D80, and I see no reason why it should. Same sensor, several years more to finetune the firmware for it. It should have been a match or slightly better.</p>

<p>My comment in my previous response should read:</p>

<p>"....I'd hope people would think twice about blindly recommending the D3000 (or a D40x/D60) to newbies simply because they are newer." Not D300, as I typed in error.</p>

<p>Joe, I fixed your typo by adding that 0.</p>

<p>On the other hand, I wonder how many casual photographers really need ISO 1600 performance or a little more noise will totally ruin their 4x6 print or small JPEG they share with friends. I still prefer the D3000 over the D40 and D60 because of the improved AF and the larger LCD on the back. Now that I am very used to the 3" LCD on the D300, it is hard to look at the smaller ones on the D200 and D2X.</p>

<p>The CAM 1000 has me holding out on a final desicion, but the image quality of the D40 is a strong force. I posted ISO 1600 because that it what was requested in the other thread. The fact remains, as I have said, that even at ISO 200, I see noise in D3000 images (while the D40 is perfectly clean). Will it show up in 4x6s? Probably not. But 8x10s <em>I've printed </em>look different. And that's kinda silly at ISO 200 in 2009. It is a terrible letdown on the D3000.</p>

<p>AN UPDATE TO MY RESPONSE ABOVE TO RADEN</p>

<blockquote>

<p >Joe A, Oct 08, 2009; 09:30 a.m.<br>

Raden.... If you think the D40 is slow, you should see the D3000. Even shooting JPEGs, it locks up for 2-3 seconds after a single shot with NR on. Can't bring up the shooting display to take the next shot into the buffer. Can't open the menus. Can't play an earlier shot in preview. Nothing. The camera is dead. Does the same thing with NR off after a 3 or 4 shot burst.</p>

</blockquote>

<p>I did have a problem that caused the delay/lockup. Turns out the battery was the culprit. I was able to try another battery and the "no response" lockup problem went away.</p>