Sigma 50mm 1.4 Fringing question

<p>I just received my sigma 50mm 1.4. I was a little bit skeptical to buy this lens because of many reviews about front and back focusing problem. But because of so many praises about the lens, I decided to get one. I bought it used and just got it yesterday.<br />I took some pictures with it. Nice bokeh like everybody says and thankfully, no front or back focusing problem. I was so happy until I took a picture of a squirrel on a tree and the ears became purple including so many of the tree branches. The background was bright sky. I was shock. I was so shock that I deleted the picture immediately.<br />I did some research, none of the reviews says anything about severe CA or purple fringing. Photozone and DPreview even say that this lens is supposed to be better than other 50mm 1.4 lenses in terms of fringing and CA. So, this morning I took some pictures again and the fringing is still there and so severe.<br />My camera is Nikon D90. Picture quality is set to vivid normal. ISO 200 or Lo1 with manual mode.<br />My questions:<br />1. Is this a problem by the lens design or I got a bad sample?<br />2. Can it (I mean the lens) be fixed? I know the pictures can be corrected using some softwares but I don't want to do it all the time if other lenses of this same model do not have this kind of problem. Should I send it to Sigma to be fixed? <br />3. Is Sigma's warranty transferrable? I have the receipt and the lens is purchased from B&H.<br />4. Anybody has similar problem?<br />5. Should I buy the Nikon 50mm 1.4G?<br />Thank you everyone.</p>

<p>I apologize for the other posting. I did not know how to ad picture, so I posted this one with picture and I could not delete the other posting.</p>

<p>Moderator, please delete the other posting. Thank you.</p>

<p>If the picture is a crop, and is not resized, I think the amount of LCA is normal. I have the Nikon 50 1.4 G and I would expect it to behave about the same way. The new version of NX2 support a new feature that reduce LCA, it works quite well, but you have to shot raw.</p>

<p> </p>

<p>I shot raw and saved it using aperture in jpg small size so I can upload it here. No crop no modification. Here is another example. I went to the squirrel tree again this afternoon and took similar shot.</p><div></div>

<p>I have the same lens and I swear this is the sharpest lens I ever used in my life. From my experience, its even sharper than the Canon 50mm f1.2. Focus is little slow even with HSM though. I love the lens very much.<br>

Just from the picture you take, I can't tell if its the problem of the lens or any lens or the sensor. From all reviews, there is no different optically between 50mm 1.4D or 1.4G. If you just want to test it out, you could test the 1.4D. Again, optically, its no way either Nikon 50mm 1.4D or G can compare with Sigma 50mm f1.4.</p>

<p>I have the same lens and I swear this is the sharpest lens I ever used in my life. From my experience, its even sharper than the Canon 50mm f1.2. Focus is little slow even with HSM though. I love the lens very much.<br>

Just from the picture you take, I can't tell if its the problem of the lens or any lens or the sensor. From all reviews, there is no different optically between 50mm 1.4D or 1.4G. If you just want to test it out, you could test the 1.4D. Again, optically, its no way either Nikon 50mm 1.4D or G can compare with Sigma 50mm f1.4.</p>

<p>Do you have this problem with your lens Eric?</p>

<p>I never really tried the lens with such white bright sky. All my pictures with blue skies, I don't see that problem. I have to try and see. If I found the same problem. I will still love the lens but I have to use the lens more carefully.</p>

<p>The 50/.4AFS certainly shows the same effect at wider apertures. Of course is more notable in this kind of backlight situations. I didn`t tried yet that correction feature in NX2.</p>

Oh dear! I just responded to the other posting. Here it is again:

<blockquote>

<p>

CA comes in two flavors, lateral and longitudinal.

</p><p>

Lateral CA comes from the fact that the focused image has different sizes for different wavelengths. You can easily correct it in any RAW converter, and for JPEGs your D90 does it automatically. You see lateral CA as fringes around twigs that have different colors on different sides of the twigs, one side green, the other magenta, but blue/yellow may also be seen. Lateral CA are no problem in the center, but increase towards the corners.

<p>

Longitudinal CA can't be corrected in software. They come from the fact that the focus for different wavelengths lies on different planes, i.e. when green is focused on the sensor, red is focused in front of the sensor and blue behind. Or vice-versa. Does not matter here. Longitudinal CA are everywhere in the image, not only near the corners. They can be seen as green fringes around out-of-focus background objects and magenta fringes around out-of-focus foreground objects. The fringes have the same color on all sides, they only depend on the distance, the amount by which the object is out of focus and the aperture. The bigger the aperture (the smaller the number), the stronger are the longitudinal CA.

<p>

It is hard and expensive to correct a lens for both kinds of CA. Lens makers don't seem to care much about lateral CA today. the cameras begin to correct it automatically and it is not more than a slight nuisance. They normally concentrate on longitudinal CA, but even that is hard to correct for a lens used wide open. The faster the lens (i.e. the bigger the aperture), the worse the problem. If you look at tests of the very expensive and highly regarded Canon 50/1.2L, you will see that it has much stronger longitudinal CA than the Sigma 50/1.4. That's just the way it is. You can spend $4000 or more on Leica lenses that are better in that regard, but that's about the only thing that you can do.

<p>

In fact the Sigma 50/1.4 has significantly less of both kinds of CA than the new Nikon 35/1.8, about the same lateral CA as the new Nikon 50/1.4G and significantly less longitudinal CA than the Nikon 50/1.4G.

<p>

This leaves purple fringing. That is a different phenomenon, and it is most apparent on highlight edges, normally only on one side of twigs, it is strongest in the corners, and it is connected to overexposure. If you correctly expose for the sky, it should go away. It will also go away if you stop down. It's the combination of wide open lens, overexposure and contrast edges that triggers it. It is also not only a lens problem, it seems to be connected with digital sensors as well. In fact it does not happen with film.

<p>

If you post a crop from the corner, we probably could diagnose the problem, but if it is purple fringing, it just is that way under those circumstances. Avoid the circumstances and you avoid the problem.

<p>

I have no particular experience with purple fringing on my Sigma 50/1.4, but I am sure with the right combination I could trigger it as well. It is much less than in older Nikon lenses like my AF 24/2.8 or AF 85/1.8D, but it may be there.

<p>

Now specifically to your questions: No, it's physics. The design of the Sigma 50/1.4 is class-leading. You may get better lenses, but not even remotely in that price category.

<p>

Lateral CA can be fixed in software, longitudinal can not. In case of severe purple fringing, I normally use a Hue/Saturation layer with a mask to selectively desaturate blue and magenta where it is strongest. That's about the only thing that is possible.

<p>

No idea about the warranty, but it does not seem to be a case of a defective lens (though an image would help to verify it).

<p>

I suppose I'd have a similar problem under similar conditions, and I suppose I'd have it with pretty all my lenses that open up that wide.

<p>

The Nikon 50/1.4G won't help you a bit.

</blockquote>

<p>

Now that I see this image, it is clearly longitudinal CA. Some twigs in the background are green, some in the foreground are magenta. It looks a bit strong to me, but that may be due to the overexposure as well. I have not tried the last version of Capture NX2, but I guess they can't do much more than automatically desaturate along highlight edges. It may work well, but I doubt that it completely goes away.

<p>

I'd check with my own lens, but I can't at the moment. It rains outside, it is night, and my lens is 350km away from here :)

<p>

Simply forget it. Make other images or stop down to f4. You won't find a lens that looks good in that particular situation.

<p>Andreas, sorry for the other posting and thanks for the information. It's really useful. I'm glad that you and Eric and Jose all think that this is physic and not a defective lens. I know the situation is extreme with very bright sky background. I will try to be more careful with the background choice and aperture although the point of purchasing 1.4 lens is to use it at 1.4.<br>

I plan to use this lens for portrait indoor and outdoor. I will try it and see what happens.<br>

Here is the picture from the cropped corner.</p><div></div>

<p>Center cropped.</p><div></div>

<p>By the way, I have no doubt that this lens is very sharp with beautiful bokeh (bokeh example is not in this picture). The color is warm. Other than the above problem, I really love this lens and really want to keep it. That's why I want to make sure that my problem is not caused by a defect with my lens.<br>

This picture is taken at F/5.6, no post processing except for saving in smaller jpg file.</p><div></div>

<p>Are these removable in NX2 or PS/CS3(4) ?</p>

 

<p>Bare tree branches against an overcast sky is a tough test for any lens. Some of the best lenses I've owned will show purplish fringing in that situation, including an otherwise excellent zoom on my old Olympus P&S digicam. So does my 28/3.5 PC Nikkor on the D2H.</p>

<p>If the lens is satisfactory in other ways you might consider keeping it and just avoid taking photos of bare tree limbs against overcast skies.</p>

<p>Corel Paint Shop Pro X has two quick fix tools for both kinds of CA, the usual kind associated with lenses on film or digital, and the purplish fringing that plagues digital specifically.</p>

<p>I think I will keep it Lex because I really like it. Thanks for the comparison. I was taking picture of a squirrel yesterday and I got this purple tree branches result. I deleted the file but I took the above picture, the same tree but without the squirrel. Those lenses of yours, are the fringing this obvious?<br>

Arthur, I use mac aperture. I just go to the color and reduce the blue or green or purple saturation to -70 or -80. It helps. But the whole picture is affected.</p>

<p>Andreas wrote:<br />"I have not tried the last version of Capture NX2, but I guess they can't do much more than automatically desaturate along highlight edges. It may work well, but I doubt that it completely goes away."<br />You are right, that's the way Capture NX2 try to correct LCA, it makes a quite good job but sometimes it desaturate areas it should not. It's a pity that you can not mask the correction.</p>

<p> </p>

Thought so. On the other hand, there is simply no way to do it in software. It's all over the image, it is only more apparent in some places than in others.

<p>Here's an oldie from 2002, with my Olympus C-3040Z digicam. This was the only time I was able to force really obvious purple fringing, around a light bulb and edges of white venetian blinds.</p>

<p>I recall some debate about whether CA was influenced by stopping down, so I threw together this test for demo purposes. At the time some folks were referring to this specific type of purplish fringing as "sensor blooming," but that term may be out of favor now, I'm not sure. It seems to be specific to digital sensors as I don't recall ever seeing this particular type of purplish fringing on film. Unfortunately I don't have the original version testing the lens at f/1.8, where the problem was more obvious.</p>

<p>Otherwise, this Zuiko zoom was outstanding, especially for closeups and macros. I even used it to copy some color slides and the quality rivaled my Minolta film scanner.</p><div></div>

<p>Another recycled photo from last year, demonstrating the same purplish fringing using my 50/2 AI Nikkor on the D2H. It's apparent wide open but clears up quickly stopped down.</p><div></div>

<p>Ferdinand,<br>

The artifacts in these pictures are most likely a combination of CA and over exposure. In my experience overexposure (signal clipping) is the main problem here. If you look at the R, G and B histograms do you see clipping? When shooting this sort of subject on my D300/D200, I always set the exposure manually and I under expose 1/3 to 1 stop. This minimizes a lot of purple fringing artifacts (but not CA). For critical work I carefully check the RGB histograms on the camera back. Shooting animals in trees may not give you much time to carefully set exposures. But you should have time to quickly set the exposure compensation to insure the sky is properly digitized. <br>

While I find Nikon's metering to be very useful. The light meter on any camera can mislead us. In some cases, the meter reading is only a starting point. Take few minutes and experiment. When the sky is underexposed by 2 stops and see how much symmetrical purple fringing is present. Add light until the fringing becomes annoying. Look at the RGB histograms for these images. This data will help you understand how to expose this sort of subject in the future.<br>

I shoot in RAW and when the sky is properly digitized there is usually enought data to pull up the mid-tones and shadows by 2 stops or so (I use the fill slider in Lightroom). Digital capture has many wonderful advantages, but my D200 and D300 do not perorm well at all in the circumstances that existed in your photos when there is a even slight amount of overexposure (signal clipping).</p>

<p>Thanks for the sample Lex.<br>

William, what do you mean by clipping on the RGB histogram? Yes, I was not careful with metering and overexposed the background. I will try to under expose a little bit in that kind of situation. I am going to take outside wedding and I really need to practice with bright light with the lens. Do you think light meter will help? But above all now I know it's my mistake and not that my lens is defective. I'm happy.</p>

<p>Clipping happens when a sensor well's photon count becomes exceeds the sensor/digitization hardware's ability to measure it. When this happens the right side of the histogram will typically contain a tall spike. This spike indicates that light intensity of the brightest parts of the frame relative to the darkest parts is unknown, as the brightest areas were improperly measured (under exposure causes a spike on the left side of the histogram, that is there are many zero values). When sensor well overflow occurs, there is no way to know how much signal was lost. Reducing the exposure eliminates the spike which indicates that the brightest areas have been properly digitized.<br>

When sensor wells overflows during an exposure, the electrons generated in the overflowed wells contaminate adjacent wells. The more overexposure (the larger the spike on the right hand side of the histogram) the larger the artifacts become in the adjacent sensor wells. The result can be the purple fringing artifacts you saw.<br>

I don't think you need an external light meter. I think you need to realize that your camera's light meter is really only a staring point. Much of the time automated metering works very well. Other times it fails. This is easy to forget when camera automation is so easy to use. You are smarter than the meter.<br>

I would use the camera's spot meter mode to meter the sky near to my subject(s). I would use this as my starting exposure value. As you gain experience you can use the camera's exposure compensation control to quickly reduce the adjust (correct) the automated exposure setting as conditions dictate.<br>

You may have seen phtotgraphers using a flash outdoors at weddings on a perfectly clear sunny day. The flash adds light so that facial features do not become too dark when the frame is properly exposed for the bright sky. In some cases RAW images contain so much information the shadow areas of the subjects faces can be lightened in post processing, so a flash is not necessary to get a decent exposure of the subjects. But with a flash, the odds of getting a nice result are much higher.</p>

<p>Thanks William. Yes I will use flash outdoor for fill flash and shoot RAW for post processing if needed. It's gonna be in ME, in June, at 2 PM, no tent. So the sun will most likely be harsh. </p>