K-5 1.03 versus K20D AF tests

Gary is it possible you can do similar tests to mine? The focus chart I used is on page 18 of this pdf ...

http://focustestchart.com/focus21.pdf

I used a subject distance of 30"

EV levels can be got from this.

[ur]http://www.fredparker.com/ultexp1.htm#EXPOSURE%20FACTOR%20RELATIONSHIP%20CHART%20B[/url]

I've yet to find an online EV calc where yo can just dial in exposure values, but you can get a rough idea from that chart.

Quick question. I was calibrating one of my lens after the firmware update.

LV was spot on but when shooting through the view finder it was FF. So I make an AF adjustment on the lens of -8 and through the lens now is nearly spot on but then surely that would then kick the LV out? or doesn't the LV apply the AF adjustment? It doesn't appear so as now both are spot on....


I'll see what I can do this weekend Smeggy.

LV use a very differnt system so the AF adjustment have no effect.

Liveview looks at the picture and try to create the sharpest contrast, if you've a sharp contrast it's sharp. As far as i know the af adjustment doesn't make it look for a weaker contrast.

The Phase AF system is very different and you also have the problem of small variation in the calibration and aligment of the AF sensor and the lens AF.

Here you can read more about it and below that is contrast AF.
http://en.wikipedia.org/wiki/Phase_detection_auto_focus#Phase_detection

As I understand it LV simply hunts for the best contrast, so it is based entirely on finding maximum contrast on the sensor data. It doesn't need to know anything else.

If LV could be used to provide the AF correction that would be brilliant - you could just run a few tests and the AF data would then be 'locked' to the correct LV-derived focus distances, providing a highly accurate mapping of AF measured distance to correct distance. Would need some firmware though...

Like this? http://www.dpreview.co.uk/learn/?/Glossary/Exposure/Exposure_01.htm

Stefan and karma mechanic many thanks for confirming that. I thought I was going mad for a minute there... lol.

Thanks Mark.

I should have been clearer.
I've found a few calcs were you can select predefined values, but none were you can just dial in the values. That calc only allows whole steps and, for example, won't work for 1/5s and ISO320

Cheers anyway

These are two great pages for explainign the pros and cons of each AF system

http://graphics.stanford.edu/courses/cs178/applets/autofocusPD.html

http://graphics.stanford.edu/courses/cs178/applets/autofocusCD.html

You need flash installed in your browser

I found an excel spreadsheet on the other forum which calculates EV for you.

I follow these "K5 issues" with interest as it's going to be my next purchase but I think I'll wait a little while longer to see the consensus on the firmware success. I notice on DPreview that some owners report that it's solved their issue although I'm not sure what testing they carried out?

At least for me, as I mainly shoot landscapes, the low light/tungsten issue shouldn't be significant - it's a strange one though as it doesn't seem to affect all K5s

Simon

I'm fairly certain it affected all K5's Simon, just there were a lot of naysayers out there (blaming user error!)...

It was really visible on fast lens at low light. Slower lens hide the error because of the larger DOF.

I don't know about "naysayers" but it pays to be vigilant about the methodology of users. Some will be precise and obtain good results (that is, good in the sense of reliable) and others will vary down to the extreme where fundamental flaws are not even realised.

Add to that the increasingly widespread hacking of camera systems and the ability in later models to fine-tune focus and there could be quite a lot a variables.

I don't know what answer to this there could possibly be, but returning a camera for service people to deal with is one way of hopefully ensuring an accurate result. Otherwise, we may be better just doing our own testing until we as individuals are satisfied.

An example is the firmware upgrade. Some say it works, some say it doesn't and some say it's partially successful. It may be that there are other variables, it may be that some don't test properly, who knows....but in the end all that counts is whether or not our cameras focus accurately.

One thought is that this may never have been any different. No magazines have reported this as an issue, and it only becomes one at very low EV values. Before the internet, this would just not have circulated and it would have taken a very long time indeed for any issues to become common knolwedge. Are the same issues present in AF film cameras?

Unfortunately, waiting for film processing would mean that meaningful testing would be very laborious.

I haven't bothered testing mine for low light FF problems because I rarely shoot that sort of stuff but installed the new firmware last night and then took a couple of shots to see what's what.

f1.4/ISO 3200/1/125

First picture was with liveview using the 50mm f1.4 and have to say it was as sharp as you could wish for, have hardly used live view before and have to say I was very impressed with the focus.

Then took the same pic using the viewfinder ... what a difference, it was nothing like as sharp as the LV photo.

I paid the best part of a grand for the K-5 and have to say ... can't see the point of having such a high ISO beast if it can only be useful in good/medium light conditions (even if I rarely want to use it as such).

I'd suspect that a new firmware update would likely not factor into account the variables induced by users who have hacked or changed thier focusing..... we have numerous threads that cover customizing the focus (Smeggy you DID alter those settings correct? Did you revert them back to factory settings before applying the upgrade?) even those who physically removed the bottom of the camera to make adjustments with a screwdriver... Did the factory test for every conceivable mod a user could make before installing new firmware? not likely...

I wonder how many "as it was from the factory" users found the firmware update solved the problem?

Today I found that installing the new firmware has made a lens which was focusing perfectly with 1.02 need an adjustment of -8 under 1.03 to focus correctly.

Mike-P exactly the same with my 50mm f1.4 - LV is perfect, using the viewfinder is ff...