Early critical test observations of the D800 with the 24-120/4G VR

<p>Regarding the camera itself, I am quite pleased. Given it's potentially high resolution, the focus adjustment tolerances out of the box are quite high and require no MF tweaking of AF or AF fine tuning on any of my lenses even with the TC-14E II or TC-20E III. This was no the case with my D700 which required fine tuning on my extenders using AFS tele zooms or super tele primes. In fact, the TC-20E II or III with f/4 primes was so bad on the D700 that even with a maximum of -20 points on the AF fine tuning on my 600/4 AFS, it still need MF tweaking for tack sharpness. I can't really fault Nikon here as the D700 AF is only accurate with if the effective aperture is f/5.6 or faster. The TC-20E makes an f/4 zoom or prime effectively f/8. Fortunately, the D800 works perfectly up to f/8 with the central sensors as advertised.<br>

Comparing only sharpness (resolution) using 1954 USAF test charts, There was little or no difference between medium and large size files in JPEG using fine quality. Likewise, I saw no significant difference in resolution shooting RAW compared to the large files in fine quality JPEG.<br>

Regarding the 24-120/4, optically it tested sharper than some the critics claim at the edges and corners with the corner outperforming the edge! Stopping down to f/5.6 nicely improves the edge and corner and by f/8, it's at optimal sharpness. Of course centrally it's sharpest. No doubt there is distortion and fall off. In camera compensations for vignetting does not affect corner or edge sharpness. However, auto distortion compensation does reduce sharpness by 1 or 2 groups on the resolution charts across the frame making a small but noticeable difference at 100%. If maximum sharpness is desired, you might want to leave this setting off and post process the distortion compensation on your computer.<br>

That's all for now. </p>

 

<p>"Critical" test of a D800 and "24-120" just don't go together very well. JPEG vs. RAW? How about something a little more juicy like D700 vs D800 with 600/4 at base ISO? If I had a D800 I would not be using any auto features, and only RAW.</p>

<p>Looks like a useful test to me... How about some images, please...</p>

<p>Well, I have some pictures. Everything was captured on a D800 @ ISO 100, tripod.</p>

<p>This is the entired image.</p>

<p>At its base ISO 100, the D800 yields wonderful colors and details. The 24-120mm/f4 AF-S VR is excellent in the center, but the corners are a little weak at 24mm. The 24-70mm/f2.8 AF-S is better into the corners. I also tried the 28-70mm/f2.8 AF-S, which is also good into the corners at 28mm.</p><div></div>

<p>24-120mm/f4 Center</p><div></div>

<p>24-120mm/f4 Corner</p><div></div>

<p>Same scene with the 24-70mm/f2.8 AF-S, also @ 24mm, f8</p><div></div>

<p>I'm a little surprised by how GOOD that corner performance is, actually. I mean... really... at what print size will you actually see a problem?</p>

<p>I'll post the test images later when I get home but as I recall sharpness and contrast readings were similar to their better quality wide to tele zooms when tested on 100 ISO film. The 24-120 sharpness tends to soften towards the long end of the zoom range which has been confirmed by other reviews like DxOMark or Photozone. The focal length I used was 105 mm for my detailed testing as it seemed a little softer than the other focal lengths on preliminary testing so this is probably a worst case scenario of this lens for resolution purposes. Shooting at f/8, the center exceeded 72 lp/mm and the corner measured 56 lp/mm which seemed far better than DxOMark if you look at their measurement section on the resolution data. Testing was done at 100 ISO and the shutter was left open while exposure was done by turning the light source on and off so as not the cause degredation by mirror bounce or body vibration.</p>

Shun, justy a small question : are the "f" numbers printed in the pictures the correct ones, or the ones in the titles above/below the pictures ?

<p>Shun, you must stop posting images like this. I am drooling all over my keyboard. <GRIN></p>

<p>Beautiful colors and the meter handles the back lighting well.</p>

<p>CPM: I think that the f/8 in the image is how the exposure was made, and the f/4 and f/2.8 in the captions are to help you identify the lens.</p>

<p>Shun, Matt, off course, sorry to bother, must be sleepy after a long days work..</p>

<p>Sorry, everything was captured at f8 around 24mm. Focus was on the sculpture by Dale Chihuly, but there should be sufficient depth of field for the lower corners. The scuplture is inside the building whose picture I posted to this week's Wednesday image thread: <a href="00aE5Y">http://www.photo.net/nikon-camera-forum/00aE5Y</a></p>

<p>And if you like the image, it is mainly due to the photographer, not the D800 or lenses. :-)</p>

<p>Alan, thanks for doing your tests and posting your thoughts. It's good to read a critical analysis for a change of pace over the usual snap shots we get everyday. I frequently use JPEG Fine instead of NEF because it saves me time and effort in post, and I will conduct a similar comparison once I buy a D800 (hopefully soon ;-).</p>

<p>Also, good to hear about f8 AF performance because my longest glass set-up is the AF-S 300/4 and TC-17e combo. I will definitely see how it does as the sun goes down. For the D700, I have had pretty good results in full sun, but have to take off the TC when the light fades.</p>

<p>I want to like the 24-120/4, but the independent testers haven't been able to persuade me that it's worth the stiff price. Still, your comments are quite positive, so I hope to find a sample to test at some point in time. I wonder if this lens is subject to some sample-to-sample decentering that messes up corner performance? I've learned to look at all four corners with these wunderzooms becasue frequently you find a good corner, and then the opposite corner turns out to be a dog.</p>

<p>Anyway looking forward to any other comments you might post :-)</p>

<p> </p>

<p>My expeience with Nikkors is that their tolerances are fairly tight with their better lenses so I doubt there will be a lot of sample variation with modern production lenses but of course this isn't always the case as with a few famous Nikkors like their MF 35/2 Nikkor (AI or AIS). One reason this lens may have tested well for me is simply that the D800 has a better sensor and processor than any other Nikon digital camera in the past. A better higher resolution sensor is going to bring the best out of any lens if it's optics are fairly good to begin with. While this lens is pretty sharp, a good MF prime like my 45/2.8 AI-P will easily blow away the 24-120 @ 45 mm giving results of over 88 lp/mm centrally and about 72-80 lp/mm at the edge or corner.</p>

<p>Shun, interesting, and thanks for posting, but I have to wonder: to what extent is depth of field the limiting factor? Because the corner 100% from your test appears to be from an area somewhat behind the main subject. On the D800, the pixel pitch is 0.00489 mm (0.000192 in). As an example, using as our standard the circle of confusion equal to one pixel width, with a 28mm lens focused at 6 ft (about 1.8 m), the depth of field is about -0.5 ft to +0.6 ft (-15 cm to +18 cm). So anything outside of a fairly narrow band around the plane of focus is going to start showing a lack of sharpness / resolution due to depth of field.</p>

<p>Of course, getting pixel-sharp results on a super-high-resolution, relatively-large sensor camera like a D800 puts very high requirements on depth of field (and/or diffraction!).</p>

<p>These issues are why, IMO, lens tests need to be done with a test subject that is flat, and using some sort of greatly magnified check on the focus (either before exposure, using live view, or after, as an iterative process). Of course, if you're testing the <em>camera</em>, then then accuracy of the autofocus system is also something worthwhile to know--but somewhat separate from the camera's ultimate resolution limit.</p>

 

<p>Good points Dave. Perhaps I indeed didn't have sufficient depth of field for the corners, although I think the subject is farther than 2 meters away.</p>

<p>However, I do find the 24-70 and 28-70 to be sharper into the corners under identical situations, and the unshaprness from the 24-120 is more like corner smearing as we observe on a lot of wide-angle lenses as well as version 1 of the 70-200mm AF-S VR rather than just being out of focus. I am not exactly a test chart person. Maybe Alan can provide better examples.</p>

<p>Shun's observation of the less sharp corners at 24mm and f/8 with the 24-120 as compared to the 24-70 are supported by the photozone tests that show the same trend. They also predict that at 40-50mm, the 24-120 would close the gap or even have the "edge" (pun intended) on the 24-70. The 24-70 would then pull ahead again at longer focal lengths.</p>

<p>I believe that the characteristics of zoom lenses are so complex that it is often hard to call one better than the other - and of course sharpness is only one of the parameters that needs to be considered.</p>

<p>Sorry, I didn't mean to post misleading information. When I have a chance, I'll retest the 24-120mm/f4 AF-S at 24mm. Next time I'll use live view to focus the corners.</p>

<p>The 24-120 is my lens. I bought it along with the D7000 a year and half ago when Nikon had some $300 off package deal. That was pre-earthquake/tsunami/flood. Given that it is a 5x zoom, I expect it to be a lesser lens but I prefer the longer zoom range for casual photography.</p>

<p>Moreover, hopefully Alan has more information to share with us. I also don't mean to "steal his thread" so to speak.</p>

<p>Thanks for everyone's posts regarding the AF-S 24-120mm f/4.0 VR, vis-à-vis, the Nikon D800--these were especially helpful since I recently chose to go with the 24-120mm over the venerable 24-70mm for my short zoom event coverage. I noticed that the 24-120mm f/4 Nikkor was among the "Nikon-recommended" lenses for the D800, so I grabbed a refurb copy a few weeks ago. Also, it's been a fairly well-regarded lens overall. While I'm disappointed to see the corner sharpness fall off at 24mm, I suppose that's to be expected. At least I also have the 24mm f/1.4, which I think should fare better for portfolio shots (the 24-120mm would be primarily for event gigs).</p>

<p>As promised, I have posted the test images taken with the 24-120mm f/4G VR with my D800 here: <a href="../photodb/folder?folder_id=1029790">http://www.photo.net/photodb/folder?folder_id=1029790</a><br>

Unfortunately, it was the only way I could adequately show them in one location as I couldn't figure out how to add then to a post reply.<br>

Details are as follows: Focal length tested was 105 mm and test apertures were f/4 and f/8 at the center, edge and corner. All are JPEG in fine quality with the large (L) image size option. In-camera distortion and vignetting compensation were both turned off. Image crops are at 100%. Test charts are the standard 1954 USAF resolution charts. The only post processing manipulation to the image was with color correction and brightness to allow for easier visualization of the chart details as this is only demonstrating resolution. In reality, the corner and edge targets are darker than the center target epescially wide open. </p>

 

<p>Alan, did you focus separately for each section, using live view manual focus? Field curvature can cause oddities such as the corners appearing to outresolve the center of the edge while in effect it could simply be a question of one area being slightly (more) out of focus than the other. Photozone find the corners of the 24-120/4 not as good as the edges (on D3X). </p>

 

<p>Good point, no I didn't. Focus was strictly on the center target. At best I think it would only make a marginal difference at the edge and corner because the test distance was 17.5 feet to the center target and given that the maximum aperture is f/4, I think even critical DOF would cover the extra distance. Also, I checked my trusty 105/2.5 AIS with the same setup and resolution was quite a bit higher across the frame with little difference in results from center to corner, even at f/2.5. </p>

<p>Shun, are those corners actually in focus? Can we expect them to be in focus given that they are more distant than the main subject (which is where i assumed that you placed the plane of focus)? How about a flat plane sharpness test?</p>