AF in artificial lighting conditions.

[petervanl]

After several years of struggling with focussing with wide angle lenses I found out that my Minolta/KonicaMinolta camera's are backfocussing in artificial (ie tungstun) lighting conditions with wide angle lenses wider than 24 mm. In normal lighting everything is ok. Every time I switch on a lamp the camera's start backfocussing (even in conditions where the camera could focus ok in the available light), tungstun lamps are worse than economical types of lamps.
I did try some Sony DSLR's in several shops, most of the time these were backfocussing at wide angle in artificial light. I used the lenses that were on the cams like 18-70, 16-80, 20 2.8 etc.
IMHO this ''problem'' started upon the introduction of the Dynax 7 (filmcam) AF system. This system, or a form of this one, is still in use in the Sony DSLR camera's.

I would like to ask all of you to test your camera's in artificicial lighting conditions while using a wide angle lens.
Perhaps David Kilpatrick could test his 17-35 on the A900.
Please shoot a subject at aprox 3 meters with a wide angle, one in artificial lighting conditions and one in daylight conditions. Preferable the same scene. Please also report the type of lamp used (tungstun, TL tube etc). On the A700 please report the firmware version as I hope V4 fixes this problem.

Thanks for your efforts.

Peter

[Dr. Harout]

Interesting. And while we wait for some answers allow me to welcome you to this forum Peter.

[paul_32]

Welcome, and I'll give it a try asap.
For now, I shoot a lot with 14mm, 24mm, 15 mm fisheye. Wide angle lenses can have some limitations, that's why I normally stop down tot F8 or F11. So I use them in MF, set at 1m....

This describes a test for backfocus.
http://www.northlight-images.co.uk/arti ... tment.html" onclick="window.open(this.href);return false;

[Javelin]

I knew my A700, A200 and a canon 350D do it.

I now know my 7000 also does it. to a certain extent with my 35-70

17-35 and 35-70 and 28-135 only wide open in each case on the digitals too. my only camera that doesn;t do it in my F717 and thats probably because the hologram is always up in lower light.

[Gianluca]

Hi, I have an A700 with firmware updated to version 4, and I'm experiencing this.
The only wide angle that I have at the moment is a Tamron 17-50/2.8 and when shooting indoor I'm using it MF only just because of this.

Thanks for this post, it's good to know that I'm not alone, and now I understand why this is the only lens with BF issues...

[01af]

It's a well-known effect. It also happens under daylight illumination when trying to focus on vividly mono-coloured blue (--> front focus) or red objects (--> back focus).

-- Olaf

[braeside]

One would hope that perhaps in the future that AF systems would be able to compensate for different light temperatures?

[petervanl]

I have heard that some Canon cameras also show this AF behaviour, so I tried some of their cams.
Last day I have been playing around in a shop called Media Markt in the Netherlands, they have several DSLR's cameras of different brands lined up to try.
All cameras were equiped with lenses and powered, no memory cards.
I have tried several of these cams, Canon 40D and 50D with 17-85, Nikon D90 with 18-105, D300 with 18-135 and A A700 with 18-70. The shop was lighted by a mix of light sources, spots, tl tubes etc, all artificial lighting.
All the Canons and Nikons focussed properly in the wide angle range and tele range of the lenses, judged by zooming in on the picture kept in internal memory of the cameras (surprize to me that this was possible without a memory card, but it worked).
The A700 focussed properly in the telerange of the 18-70 lens, but was backfocussing very noticeable in the wide angle range.

This is of course no proof that every A700 has backfocus in these lighting conditions and it could be that the camera/lens suffer from backfocus in daylight as well, it would mean that the AF system should be calibrated.
Anyway, it was not assuring that only the A700 in the whole line of cameras was backfocussing.

It might be that Sony not only inherited good things of the Minolta line, also at least one bad thing came along.
I hope the A900 does better with respect to focusing in artificial light with angle lenses, anyone able to give this a try?

Peter

[Javelin]

I only noticed this behaviour in low wattage incandecent and CFL lighting. the CFL lights I'm talking about actually take the same amount of WB tweak to make the yellow go away so the light quality seems very similar. With really bright lights no matter what type it seems to work ok

[sparaxis]

It seems to be more of a problem with some lens and camera combinations than others. Perhaps it depends on the exact calibration of each component. I suspect the amount of longitudinal chromatic aberration a lens exhibits might be a factor too.

I know I went nuts when I first got my 7D. I was checking for FF/BF with a new 50mm/1.7 and I got wildly inaccurate results. Of course I was indoors and using incandescent light.

My Sigma el-cheapo 18-50mm will not focus accurately in any light on any body at 18mm. (We have a 5D, 7D, and two A700s)

On the other hand the 17-35 D KM lens focusses just fine in all light with both my 7D and A700.

My Maxxum 7 film cameras do not seem to have any issues with incandescent light either.

If a particular combination is more-or-less in focus in one light source it is possible that a change of color temperature could be just enough to make it look out of focus in another.

On the other hand we are far more critical than we used to be because we can blow everything up to 100%.

Many years ago the late lamented Herb Keppler did a series on the accuracy of autofocus. Pretty much no camera was reliably accurate to more than about 40 lp/mm on film. The Maxxum 9 was included and was a little worse than the Nikon F whatsit and EOS1 something in the comparison.

40 lp/mm is nowhere near enough for "pixel sharp" results even on a 6MP camera. Most autofocus systems of the time (I guess the mid 1990's), were designed for around this value as a good compromise between the desired accuracy and the amount of contrast available to focus with. If a higher lp/mm was chosen, many lenses would not have enough contrast to focus reliably.

This also explains why very contrasty lenses like the 16-80 zeiss can be fooled so easily. The system locks on as soon as it thinks it is good enough.

There is a good write up on phase detect autofocus in "Applied Photgraphic Optics" by Sidney F. Ray (A very expensive book by the way).

[Dr. Harout]

Nice post, Sparaxis.
And yes, you are absolutely right about 16-80, but still I'm really considering it.
As for Herbert Keppler (God bless his soul) I always read his articles with delight.

[Dr. Harout]

Sorry and shame on me, I forgot...
A big welcome to Sparaxis

[sparaxis]

Thanks for the welcome!

I have been lurking on this forum for a while while really enjoying the discussion. I have been actively photographing since the 1970's so hopefully I have some useful information to share. I started with an SRT-101 like so many others. Herb Keppler was a huge influence on me, and I suspect so many others as he had a very down-to-earth practical approach to photography. I still have a full set of Keppler's Columns filed away starting from the mid 1970's. Sometimes for fun I start to read through them and realise what it was like in the "good old days".

But enough nostalgia.
I am also very temped to buy the 16-80, concerns about possible QC issues notwithstanding.

I have the 17-35 and 28-75 as I was planning on buying the A900. Then I got to be one of those lucky ones that bought a "close-out" A700 from Best Buy. Really lucky in fact as the only one they had was the demo model (no-box) so I got an extra $100 off. I am not surprised that they did not sell - Our best buy had a batteryless body mounted on the display. How they actually expected to sell them if you could not turn it on and look through the finder is beyond me.

The urge to move on to the A900 is lessened somewhat, although I am still lugging around my Rolleiflexes (a '55 and a '59) and my Mamiya 6 for "serious" work. It would be nice to eliminate the shoot-process-scan cycle!

[Yagil Henkin]

When I focused and shot immidiately (all shots @1.4 under warm fluorecent light, Color temperature around 2500K/M4) the camera misfocused a bit more than when I waited half a second after the red focus confirmation light appeared. Same result with another lens. Am I mad or does the focus get better even after initial lock?

[petervanl]

I have tried my 17-35D on A900 in a shop lighted by a mix of tungstun and fluorescent lighting. In this condition I had to adjust the AF micro adjust to -7 to solve the backfocus in the 17 MM range. When using a wide angle setting the AF adjustment to -20 has a huge effect on the focussed distance it went from nearly infinite to 1.5 meter. For me this is an indication that the system is most picky for focus errors with wide angles, at least it shows significantly on the distance scale, in real life stopping down helps to cover the misfocussing.

Since I only could try the combination in this light source I am still curious what the difference is in focus accuracy in different kind of lightsources. (not mistaken by different levels of brightness, lower brightness will give more problems with focussing)

Regards,

Peter