Tamron 18-250mm f/3.5-6.3 XR Di II IF-LD

THE NEW Tamron 18-250mm has some importance to Alpha system users. It is almost certainly the optical basis for the forthcoming Sony SAL 18-250mm version, and as a APS-C 13.7X zoom with an equivalent 27-375mm (in full frame terms) range is uniquely suited to the SSS-enabled Dynax 5D, 7D and Sony Alpha 100. While it is certainly true that Nikon owners can now obtain an image stabilised travel lens in the form of the 18-200mm VR, it’s not got the same range as this new 18-250mm and that puts the in-body stabilisation brands yet another step ahead.

210mm, f6.3, Alpha 100

Our lens arrived the morning that I drafted this test, and in the following weeks we have found nothing to change the opinions given here. It’s not my personal lens, and lives on Shirley’s Dynax 5D. However, I’ve made a point of using it regularly. To my dismay, it threatens to blow my CZ 16-80mm out of the water and I almost dare not line the two up against each other, which I guess I will have to do for peace of mind! Why does the Tamron do so well? Easy – the CZ simply doesn’t focus as accurately.

The Tamron is the slowest focusing lens I have ever used on the Alpha 100 (second in line to the 18-200mm it has replaced). It’s awful; it is noisy and it takes ages to make its adjustment. But when it does so, it nails the focus, no matter whether at 18mm and distant or 250mm and ultra close. This applies to our 5D and A100 alike. Moreover, even at full aperture it returns a level of detail not so far off the CZ, and does so with no higher a level of chromatic aberration or intrusive distortion. What it lacks is any ability to track focus, or even grab a candid shot, while the CZ 16-80mm does that very well with near-instant snap-to autofocus.

To clarify exactly what the lens is like, I have made a studio recording talking you through its startup, focusing, and shutdown operations when attached to a Dynax 5D. This is a 3.5MB .mp3 download and is recorded using twin Behringer C2 pencil condensor mics close to the lens, with a vocal mic higher up. It’s far more revealing than, for example, a camcorder podcast would be.

Compared to the 18-200mm Tamron or the 18-200m Konica Minolta, the new 18-250mm is in a different class optically. Tamron must have learned something as it is no bigger (in our photos the awful white plastic Minolta-type ‘fall off every time’ lens rear cap gives the impression that the KM 18-200mm is shorter). It’s a touch heavier, but it also feels far more solid. The zoom will still extend a little of its own accord if suspended aimed down and shifted off 18mm, but not in a single trombonic sigh all the way to 250mm. If you don’t shake the camera around, any setting will hold firm at any orientation. Generally the zoom does not cause gravity-aided hassle.

It also has a zoom lock, which only operates at 18mm, ideal for carrying the camera free from self-extension worries. Hint to Tamron: quadruple the value of the zoom lock by making it work at all focal lengths! When extended to 250mm, the lens is appreciably longer than the 18-200mm gets:

It’s amazing that Tamron have managed to get f6.3 at 250mm in to the same front element diameter and filter size as 200mm f6.3. You might think that the extra focal length would lead to a different lens shade, but of course, that is determined by the 18mm setting only. It therefore uses exactly the same shade as the 18-200mm, and the Tamron and KM designs look identical. Maybe the 18-250mm shade has a touch more curve on those petal edges.

So, we have a lens which pretty nearly matches the unbeatable CZ 16-80mm for rendering fine detail, mainly because it is a good optical design which has unusually accurate, slow-operating focus. It does not zip back and forth, but creeps up to the focus lock point. It beats the 18-200mm variants optically by a long way, and appears to be mechanically upgraded. Hopefully the Sony equivalent will be equally good, but they will probably omit the zoom lock as this is not part of the lens-range design. KM omitted the zoom lock on all their Tamron derivations, and a bad decision was made there. Sony have inherited that decision. It is not too late to change.

As for the things which really matter, it takes the same 62mm filters as the 18-200mm and many other Min/Sony designs, which means it can inherit a UV if you are upgrading. All my shots were taken with a Rodenstock UV filter fitted, even my into-the-sun flare test. This is real world, not test lab. I use a UV filter all the time, so I want to know if the lens creates reflection havoc with one fitted. It does not. As you can see, even in the shot with the sun right in the picture, we just get a predictable string of lens-element reflection patches and a generally radial flare from the filter. There is barely any overall veiling glare. With the sun out of picture, the combination of the petal hood and the effective multicoatings makes for a flare-free shot.

Distortion has been a major worry with superzooms. Tamron, from the first 28-200mm right through to the 28-300mms and the digital 18-200mm, has always balanced the geometric distortion so that the wide-angle angle suffered a bit of barrelling, but pincushion was well controlled at the long end. Sigma, in contrast, prefer to make the best possible wide-angle geometry and have a party with the cushions from standard to tele lengths – a real old pillow-fight, which can knock the stuffing out of your long-lens architectural details. Well, with this one, Tamron has changed. They have aimed for better wide-angle geometry (optimised, like the CZ – it’s not really that great but it will give straight enough lines next to the edges of your shot, especially on horizontal compositions like room interiors) and somehow managed a half decent tele end result. It’s still bendy, but only needs Photoshop or DxO type correction for critical work. The example above is at 250mm. It’s pretty odd to the eye. For subjects like this, you really need a very low distortion lens or some post-production correction.

This vertical 18mm shot actually shows more visible bending than a similar horizontal. It is my standard test subject for wide-angle lenses, and this is not a bad result. I have blown up a small section to show chromatic aberration or colour fringeing:

This also gives you an idea of the native sharpness at 10 megapixels. All my images here are exported from raw Sony A100 .ARW files using Photoshop CS2. Later on, you will see some examples using 17 megapixel export to 5120 pixels image width. This really tests the fine detail recording ability of a lens+camera combination. The CA shown above can be corrected, and again, you’ll see a detailed example at the end of this article.

Vignetting has always been a bit strong on the Tamron superzooms. Even Carl Zeiss, in their leaflet with the 16-80mm, explain that it is inevitable at 16mm. Again, this new Tamron zoom overcomes most previous problems and produces what must be as close to state of the art illumination at all focal lengths. I tried it with some subjects designed to show it up badly, though not as badly as an underexposed shot on a dull day with a white sky. Since I don’t take underexposed shots on dull days with white skies, but stay at home and write articles instead, I prefer to ignore the worst case!

Sky vignetting at 250mm, f8 – remember, this is only half a stop down from maximum aperture.

A light subject, minimal vignetting shown this time at 250mm and full f6.3 aperture. This is a much more acceptable result of tele-end distortion, too. You may also like to see an autofocus image detail from this shot, again this is a 100 per cent section of the original 10 megapixel image – just consider the tiny fine black low contrast lines resolved on the bottom transom of the window-sash where the paint has cracked – this really is single pixel detail, something the Alpha 100 does better than any other camera made:

Secondly, an example below to show vignetting at 18mm, using a sensible f8 aperture:

And, once more, a section taken from this shot to full size 100 per cent clip of the 10 megapixel file:

This time, now you’ve stopped looking at it, I will admit I am fooling you. This is not a 100 per cent clip of 10 megapixels. It’s actually a 100 per cent clip of 17 megapixels, the full image from this export would be 71 inches wide on a 72dpi monitor. Copy this sRGB image, change it to 240dpi without resampling, and print it on a spare bit of paper. You will be looking at part of a 14 x 21 inch inkjet print. There’s a little excess noise as the A100 underexposed by a full stop at 18mm despite the ‘standard’ nature of the subject. That is not noise under the door-arch, it is the A100’s excellent dynamic range recording a metal grille. This image does not have the ACR sharpening I generally apply (25) and it also has no USM for web display. It is as soft as possible. There is a reason for this – any sharpening actually destroys detail. If you want to see it with sharpening applied, copy the JPEG, and apply different levels of sharpening yourself.

To confirm the levels of detail and accuracy of focusing, I tried a quick test at 18mm and closest possible focus (0.45 metres, or 1.48 feet):

Here’s a close-up from the focused point – and again, I have taken this beyond the native camera resolution to 17 megapixels. There is maybe a hint of subject movement as it was windy day, but look at the individual pollen specks on the petals (top right of the clip especially):

To check 250mm wide open, as you are most likely to end up shooting wide open at the long end, I picked a difficult subject – foliage with branches and twigs against the light. This is destined to show up CA and purple fringes, and it’s also a digital-unfriendly sort of texture:

Now we will take a central part of this, and blow it up again to 5120 pixels wide via Adobe Camera Raw interpolated raw conversion:

Well, it’s not perfect, but this is an 18-250mm zoom, at 250mm wide open at f6.3, autofocused. And this is, remember, a six-foot image in full on a 72dpi monitor. So to make things worse, I will go right into one extreme top corner and see how bad the chromatic aberration or colour fringeing becomes:

This is not so hot, but remember, it is all magnified here compared to a normal 10 megapixel JPEG. The correction necessary at this setting is shown below – each lens will need a slightly different setting and sometimes it will vary for each picture, depending on focus. My UV filter could also be adding to this, it’s a risk you run if you want to protect your lens. Here is Photoshop ACR raw conversion with CA corrections applied:

It’s not perfect and here the Tamron differs from the Carl Zeiss 16-80mm, which also has strong CA. The Zeiss CA is very sharply defined, like a second image out of register, and when the correction is applied, the image snaps into sharpness. The Tamron colour aberrations are softer and more complex, leading to an overall softness even when corrected. Even so, I reckon this is an exceptional outcome for the specification of lens involved.

As for lens apertures, the 18-250mm is f3.5 at 18mm and the moment you move to around 20mm, it’s an f4 – that one-third stop is elusive! It becomes an f4.5 at 35mm, f5 by 50mm, f5.6 at 70mm, and from 135mm to 250mm it is f6.3. You might imagine it would be similar in rate of loss of maximum aperture to the CZ 16-80mm (f4.5 at 80mm) but it’s not as fast at medium to longer lengths.

Colour just glows through this lens as you can probably tell – but it’s not quite as warm as the CZ, nor do the images have exactly the same three-dimensional quality. I think this is down to the optical design and the iris diaphragm position, and a certain extra CZ sharpness (even if too often degraded by focus inaccuracies). The CZ 16-80mm is a masterpiece in terms of bokeh and the balance of microcontrast with overall contrast. The Tamron is just a very nice contrasty, colourful lens.

But… it is also substantially better than the 18-200mm, and I would be very surprised if lab tests (the lines on target sort of tests) did not show it to match or better the Nikon 18-200mm VR. The performance our lens has turned in is very good news for Sony, if they are indeed using this design for their own SAL 18-250mm.

It was a good walk, only 20 minutes in all to shoot enough tests to judge where this lens sits relative to others I use and have tested. It’s good to live in a town which has a few subjects within a couple of hundred yards that I have got to know pretty well for judging lens results.

You can download a sample raw file from the shot of Kelso Town House taken at 18mm, and also the one of the old windows above the jeweller’s shopfront taken at 250mm, both of which have been used for close-up detail examples.

For those who can’t use a raw file, I have added (May 4) a 10 megapixel JPEG made from a very underexposed raw image, which happens to show great depth of field at f6.3, 18mm, and extremely accurate focusing combined with awesome sharpness. This is lightened from a 2-stop underexposure created by reflective lettering on one sign! I have very rarely come across any 18mm for APS-C which rendered detail like the text at the bottom of this shot as well. It’s very difficult for a wide-angle to maintain freedom from aberrations at the extremes of the field when focused close. The Tamron also shows excellent maintenance of sharpness throughout the depth of field, and a very clean rendering of the unfocused background into the corners. Here’s a small repro of the shot, taken on our election day here in Scotland:

Have look at the bottom of the lowest sign, at full 10 megapixel resolution:

And the bottom of the next placard up:

And even further info the depth of field, the top two signs:

My conclusion is that our £369 expenditure is well justified. Our KM 18-200mm should sell for a reasonable price and the gain in telephoto reach is not all Shirley gets for her ‘everyday and all day’ lens – she will also get sharper images.

And here we have Shirley shooting with the 18-250mm in June 2007. OK, I cheated and combined two RAW exports or else the chateau of the Popes at Chateauneuf-du-Pape would just be a black silhouette. Shot on the 11-18mm. And we did order a half-case of assorted C du P wines for delivery after our return!

The one failing this lens has is the focus speed. There would be little point in buying it to snap your dog racing towards you across the lawn, or a toddler doing circuits of the living room; they would be gone before the lens had even started its focus action. Since many people use long settings like 250mm for action sports shots, they will be disappointed with the 18-250mm. If you think the Sony/KM/Minolta 75-300mm kit lens is slow focusing, it’s not compared to this. Sorry!

But there has not been a better zoom made in the 10X or over category for the APS-C digital sensor format. That it happens to be the widest range zoom yet made is a bonus.

Please note that many images are saved at JPEG Level 8. Some slight loss of the original detail is present. The two images associated with the raw files downloadable at the end of the article are at Level 12. Web-specific unsharp masking has been used on most images but not on the 17 megapixel enlargement from the 18mm view, which has zero sharpening either in ACR or Photoshop. It was proving destructive of detail which is present.

– David Kilpatrick, Maxwell Place, Kelso, Scotland – May 2nd 2007. All photographs © David Kilpatrick/Icon Publications Ltd. No part of this review may be reproduced without permission but you are free to quote individual comments, or link to images, so long as the URL is made visible and quoted text is properly acknowleged. A printed review based on longer-term use will appear in the Summer 2007 edition of Photoworld magazine.