Proof of concept regarding self coding of lenses

[osera]

Sean-
This has been discussed here before, I think, but after your testing and thinking so for, do you have any feel for the level of sophistication of the "correction" for vignetting that the camera is applying to the raw files? Software like DXO provides for correction of vignetting in what they say is a very camera and lens specific way (although not for Leica).

The DXO site describes what they say is the complexity of vignette correction.
Vignetting

I've always heard that corrections like sharpening and contrast are better applied to the raw files on the computer, rather than relying on more limited in camera algorithms. I wonder if all these in-camera corrections for luminousity and cyan vignetting will eventually be better provided by in-computer software. And will there be a big enough market of M8 users to make development of such algorithms viable?

[mike prevette]

Allen,

The M8 doesn't correct for lens correction, it corrects for the sensors vignetting.

I know it's a weird concept, but the vignetting caused by the sensor is a known based on the exit pupil (that's basically the point at which the light exits the lens) From that point you can calculate the angle that the light rays are hitting the sensor, and there for the amount of vignetting it will create.

_mike

[wparsonsgisnet]

Originally Posted by sean_reid

The light streaking, etc. problems are fixed with a hardware, not firmware, change. Sean

Thanks, Sean. I am under the impression that a circuit board, but not the sensor is being changed. I should have been more precise. Firmware would be software in the camera, whether encaptsulated or not; hardware is a board change.

Have I said I am ready to get my camera?

Regards,

[sean_reid]

Originally Posted by osera

Sean-
This has been discussed here before, I think, but after your testing and thinking so for, do you have any feel for the level of sophistication of the "correction" for vignetting that the camera is applying to the raw files? Software like DXO provides for correction of vignetting in what they say is a very camera and lens specific way (although not for Leica).

The DXO site describes what they say is the complexity of vignette correction.
Vignetting

I've always heard that corrections like sharpening and contrast are better applied to the raw files on the computer, rather than relying on more limited in camera algorithms. I wonder if all these in-camera corrections for luminousity and cyan vignetting will eventually be better provided by in-computer software. And will there be a big enough market of M8 users to make development of such algorithms viable?

Hi Allan,

I don't. But I can say that the tests I did of the correction's effectiveness for "artifical vignetting" (M8 Review Part 3) showed that it worked quite well. Over the next few days as time and, especially, weather allow I'll be doing white wall and real world tests of the newly coded 28s with and without the 486 filters. That will take some time. Next up would be either the 24/25s or the 21s (each as whole reviews) including this aspect. As always with my reviews, there won't be elaborate numerical analysis. I'm not an engineer nor do I pretend to be one. But I will get a sense of how these things shake out in "real world" use. It takes time with the lenses to be able to say much that could be meaningful about them. I can say right off, however, that the coding reduces the appearance of vignetting with a lens like the CV 28/3.5.

Cheers,

Sean

[carstenw]

Thanks to friendly co-M-lens owners, the lens code table has been updated with more pictures of codes, including the 90 Macro and its adapter.

[carstenw]

I have added a credit to Popflash, who have agreed to let me use their pictures for the codes. I am still missing pictures of codes of two lenses which ought to be easy:

- the new Tri-Elmar (does anyone have one yet?)
- the 90/2 Apo-Cron.

[guy_mancuso]

Carsten excellent work on this and for all that have worked the issue this is just fabulous. What team effort going on around here to get folks going. Hope you get your M8 back soon from repair and my 135 mm shows up soon at your door. I need to work on my 24 and 35 lux for this

[carstenw]

Thanks a lot, Guy! I figure it is my turn to give back to the community; I have received so much. Anyway, it is kinda fun.

A new table is up, with a photo of the 28mm Elmarit-M f/2.8 IV, thanks to Vincent.

[ponsoldt]

Has anyone tested whether the coding makes a difference in the photos. Maybe taking a shot with a coded vs an uncoded lens of the same type?

[carstenw]

Sean is currently doing such testing. RobSteve has done similar tests on the DMR, and it does make a difference. You don't need the uncoded lens; you can turn the feature off in firmware.

[carstenw]

Thanks to Paul Steunebrink I have just been able to add codes and pictures for the Tele-Elmarit 90mm f/2.8 II and the Apo-90mm f/2.0 Asph to the Leica Lens Code table in my sig.

[jlancasterd]

Originally Posted by mike prevette

With nothing more than a sharpie I was able to trick my M8 into thinking my Zeiss 21 was really a 50mm Leica lens.

[url=http://www.flickr.com/photos/freedomfromgravity/]

For the edification of those of us in UK and elswhere - what exactly is a 'sharpie'?

[scott kirkpatrick]

It's a permanent marker with a strong black and a relatively sharp point.

scott

[carstenw]

For the Leica Lens Code table in my link:

I just realised that I had fallen behind wrt. "equivalent" codings. I also moved the numerical value next to the code, and added a credit to Mark Norton who actually provided a lot of the images.

[M8-]

Gentlemen

Really need your help on this DIY coding. I have tried putting markings on the lens directly or have a piece of paper with marking cover the sensor.
But it does not seems to work.
I have turned the Lens Detection ON. But when I play the image it just does not have the lens inform.
Any trick or tip I have to pay attention? My firmware version is 1.092. Any issue with that. Or something wrong with my M8.

[V64]

Carsten,

As someone who has only 'succeded' in coding a 35mm V4 'cron (which has a screw head where the black marks are put), I have some questions/observations which maybe relevant!

The 6-bit LED/Sensor strip has what appears to be a 'wall' between the sensor and the emitters (which is clearly needed) - so there has to be some Clearance between thie top of this wall and the lens mount if any IR from an emitter is to be reflected onto it's sensor.

So the 'coding pockets' would appear to serve two purposes - to ensure that any white/black marks will not be worn away, and to provide the needed clearance over the sensor for the IR to 'get over the wall'.

Whilst the 'lateral position' of the sensor strip could vary without causing any problems, the vertical tolerance variation may be such that on some cameras a 'high position' would reult in the 'wall'being very close to the surface of the lens mount, and this would prevent any IR reflection from a lens mount without the 6-bit pockets.

So Pockets and Marks may both be needed.

Does anyone know if IR detection is enabled by the 'Black' or the 'White' marks? That is, does Black reflect IR or is it the White that does so?

Or to put it another way - is a positive IR detection (at the hardware sensor level) read as a 'zero' or as a 'one' by the firmware?

David H.

[LJL]

David,
Not going to try to answer all of your questions, but I have finally successfully coded a CV 15/4.5 to appear as a WATE. First thing was for me to get the older Leica 9cm LTM mount and swap it with the CV mount. The reason was the flat surface (no cutout) near the sensor. I wanted to test out some enamel paint (black and white) for both durability and effectiveness. Not sure on the former, but the latter works. I found it really is pretty important to get the black bars in the correct position. I tried 3-4 times with painting a black bar for the WATE code. both position and width seem to matter a lot. This is where the pits or wells may actually help, as the slightly greater distance (fractions of a mm?) seem to be more forgiving for the sensor.

My point is that the technique does work, and one can use an enamel paint, though it will tend to peel off, so care must be taken. I painted as thin a coat as possilbe, covering a larger area for the white, then applying a black bar on top of the white. So far, it is working, and the M8 sees the CV 15 as a coded lens, defautling to 18mm.

I am about to tackle my Zeiss 25/2.8 Biogon ZM next. First thing was for me to order a different lens mount from Zeiss. The 25 ships with a lens mount that brings up the 28/90 framelines, so I am replacing it with one that brings up the 24/35 framelines. I think I will only need black paint, since the mount is a very shiny silver to start, whereas the older Leica mount on the CV was dull, dark brass that did not relfect much.

I am also playing around with a CV 50/1.5 Nokton and a CV 35/1.2 Nokton to get coding on them. The Sharpie marker technique has not worked well on these for me to this point, but it could be because of improper placement or width of the black bars. Not sure, but so far it has not been successful.

Bottom line....I think you are correct in your quesitons about proximity of lens mount to sensor. However, I have been able to get things to work without wells/pits, on a very flush mount, but the bar placement has to be spot on and not too wide or too narrow. Just my observations to this point.

LJ

[guy_mancuso]

LJ I am wandering if clear nail polish will protect that code. That stuff stays hard for a long time or even clear epoxy

[Philippe D.]

Originally Posted by guy_mancuso

LJ I am wandering if clear nail polish will protect that code. That stuff stays hard for a long time or even clear epoxy

I think all that stuff is much too thick.
It will not last long, quite shorter life then the bare marker spots, whitch are 1/250 mm think only.
That why it as been made in a hollow place by Leica. They just cogitate hard before doing it.

[bill vann]

i've done very well with hand coding via sharpie with no real problems, usually on the first try.

have a 28 ultron that shows correct coding in the exif,

a 35 ultron (for sale), you can look at the pic on RFF you'll also see i used a complete black bar covering the multiple black dots needed. i'm using a correct 35mm adapter

i also coded a 21 skopar with a leitz adapter prior to selling it as i scored a 21 asph at a great deal. am now using this on a cv15, the leitz 28-90s are tough to find the 35 worse.

didn't bother with my 50mm Nokton or 90 Apo Lanthar (also for sale) and didn't try the 35mm color skopar (also for sale) but since the plat does not have a cutout i'm sure it would do fine.

bill