CHDK firmware hack discussion (14)

This is a sequel to the old threads discussing the CHDK firmware.
The CHDK fw is a hacked firmware for DIGIC II cameras. Currently
supported cams are A610, A620, A630, A640, A710 IS, S2 IS and S3 IS
(see the FAQ for a complete list).
The CHDK fw implements new features like RAW, battery indicator,
live histogram (RGB/luminance), zebra mode (blinking areas of
over- underexposure), DOF calculator, scripts (intervalometer,
exposure/focus ... bracketing etc.), text reader, file browser,
calendar, games and much more. New features get implemented
frequently, so check the timeline for the latest changes, or the
FAQ for a complete list.
The CHDK firmware does not touch the original firmware of your
camera. It is an additional program which gets loaded into the
memory of the cam. If you have a problem, just turn your cam off or
remove the batteries, and the CHDK fw is gone.

For further information, a guide, download links and FAQ see the
english WIKI:
http://scratchpad.wikia.com/wiki/CHDK

Timeline:
http://tools.assembla.com/chdk/timeline
Dear Vitalyb, GrAnd and all the others: Thank you so much for
creating this wonderful piece of software!
Some screenshots:

Main menu

Histogram (Blended) and Zebra overexposure indicator

Script menu

Link to the previous thread:
http://forums.dpreview.com/forums/readflat.asp?forum=1010&thread=23816681

Hi Vit40,

I'm very intereting by the program you made to configure DNG4PS.

I own a S2 and maybe it's the same process. For a long time i'm lokking for a good profile for my Camera but i can't find one.

The program with a small explanation will be very nice of you.

Thanks.

http://scratchpad.wikia.com/wiki/CHDK/uBASIC/Scripts:_Tele-Macro-Mode

Has anyone tried this on any other cam beside the S3 IS?
Does it work? Does anyone have any sample images taken with it?
It sounds very promising, if we could find a safe way to implement it.

So, for DNG tweakers, here's my version of A620 DNG header, along with a program to set or change DNG color matrix. Check readme for explanations

I did this about a month ago, hope I packed the latest versions

Use on your own responsibility

http://www.MegaShare.com/243293

Would it be technically possible for CHDK to implement a (very crude) motion detection algorithm using the viewfinder or continuous exposure data as input?

If no, what technical reason is preventing this?

If yes, what would be the difficulties to be overcome? What kind of trigger algorithm might reasonably be implemented? Is it possible to monitor some auto exposure value every second, and if the change from previous second is greater than X, trigger the shutter?

alvevind wrote:

Is it possible to
monitor some auto exposure value every second, and if the change from
previous second is greater than X, trigger the shutter?

Or might this already be possible with the current version of CHDK using a custom uBASIC script? What parameters might be utilized? Would each "motion sample" require a halfpress action? Anyone tempted to write a pseudoscript suggesting how it might be implemented..?

PS: 1 sample pr second might be a bit much. Once every 10 seconds might be more practical.

Vit40 wrote:

So, for DNG tweakers, here's my version of A620 DNG header, along
with a program to set or change DNG color matrix. Check readme for
explanations

I did this about a month ago, hope I packed the latest versions

Use on your own responsibility

http://www.MegaShare.com/243293

Hi Vit40, i'm very interested by the program to change DNG color matrix. May you send me it?
This is my mail: castaneda363@libero.it

Just use above link, click on big button FREE on lower part of the page, wait until 10 second wait period expire, then click Download. Inside zip file you'll find the program and header file

It's a command line program, no fancy interface, but it does what it should

I was playing with the zoom during video script, and apparently it works with focus as well. But if I have set_focus +1, it's too big a step. But if I use +0.1, I get a uBasic error and my camera shuts off. How do I get smaller focus steps?
And are there any other options for video, like exposure or something similar?

Also, the zoom during video script really uses 'steps', but I'd rather have a smooth zoom in and out. Is this possible?

Just tried it with much fear, after firmware acknowledged ,could not find how to select raw and battery display showing steady fall from fully charged to 65% in no time, now it is decision time.
--
jasm

On my A640, I noticed the same effect as you mentioned concerning battery display--but the indicated battery reserve did seem to improve (or stabilize) once back in the review mode. Also, once I decided which UI gadgets I could live without, and eliminated a few of those screen objects, the indicated battery consumption appeared to be more reasonable. I decided that I didn't need the full-time live histogram, and it REALLY improved the power reserve.

After, with the hack, we are ADDING to the processing overhead--so it does make sense that we are now using the display and computational processing somewhat outside of the "design parameters" a bit (regarding the amount of power, at least).

BTW, I am really happy with the RAW functionality now. It works great!

~ Jim

how do you switch to raw and which download did you select.
--
jasm

Yestarday has been released Ufraw 0.12, now canon A710 IS is supported.

Since I have the A640, I downloaded the "pre14-a640-100b-144" zip file created by GrAnd, the russian developer, I believe.

To Select Raw:

Powered up in review mode.
Pressed "print" button, then "menu" button
Selected Firmware Update (last menu entry) to load the hack
Then...
Press "print" button, then "menu" button again to get to the Hack "Main Menu"...
Press down arrow to the "1st menu item: "RAW parameters"...
Press "middle button to select...
Then select 1st item in Raw Menu, which is "Save RAW".

~ Jim

Hi vit40,

I have downloaded your program and red the readme.txt.

Is it possible to have more informations about how to configure profile settings.

Thanks

-On the A630, changing the profile in dng file with your tool and set your color matrix from the A620 gives quite better color rendition even if it has not been calibrated for it. So thanks for the tool and the matrix.

-When flash is used, which color matrix do I have to use between the two you provide ?

-Is there any cheaper chart to use than the ColorChecker ? (90€ for the chart against 220€ for the A630)

You need some basic knowledge about color spaces (in short described here http://en.wikipedia.org/wiki/CIE_1931_color_space ) and a reference camera, supported by Adobe Camera Raw. And some patience (a day or two)

The whole thing is about adjusting 9 parameters to get the same result as with reference camera. Digicam sensor has 3 primary colors, and each is defined by 3 coordinates. In DNG and dcraw, primary colors are defined in XYZ color space which is hard to visualize, so my program also accepts coordinates in xyY space, where xy conforms to first diagram on above page (CIE xy chromaticity diagram), and Y is intensity of a color. Parameters are in this order

Red_x Red_y Red_Y Green_x Green y Green Y Blue_x Green y Green Y

You need to take a picture of some reference image like Macbeth, displayed on the computer monitor. It's not important how well your monitor is calibrated, since displayed colors are not used as reference, but picture of them taken by reference camera. You need to take a similarly exposed picture, with histogram to the right. In my case, it showed up that with same aperture, I need shorter exposure with A620 set on ISO 50 than with 400D with ISO 100

Then, you develop a raw from both cameras with same setting (exposure, WB, etc)

With Coffin's profile, A620 WB is quite wrong, needing about 6500K and tint about 20 for typical daylight, while 5200K with tint around 0 is expected

Since sensors for digicams are similar, you can start applying a A620_V17.cmd to a DNG (on the prompt, type A620_V17.cmd Yourfile.dng), taken with your camera and see what you get. First, you must set Y parameter in the batch file for red and blue to get correct WB. Increasing absolute value of Y increases red/blue intensity in the developed picture

After setting Y parameters correctly, which will take 5-6 attempts, you should check hue and saturation of all colors and move R, G and B in xy plane accordingly. This will take much longer time, since on every iteration, you need to set WB again. In general, moving a point further from the center of diagram will increase saturation around that color, and moving a point in counterclockwise direction will increase hue

Note that, unlike for sRGB or AdobeRGB, primary colors are lying outside the diagram (except red which is inside)

When/if you are satisfied with the result, you can modify a header file of DNG4PS (in my case ...\hdr\PS_A620.bin) with the same arguments, so newly generated DNGs will have new color matrix

I'm glad that matrix works for A630, although I suppose that it would need some adjustment, as sensor is different. Read my explanation to Fragglerock

I provided only one matrix, which is calibrated to daylight, and is written also as matrix for tungsten. I was satisfied with results on tungsten lighting, so didn't bother with calibrating tungsten matrix separately. Dcraw also has only daylight matrix defined

May you provide the reference image you use for calibration, please ?

For described procedure, you can use any Macbeth image you find on the internet, like this one

Of course, you won't get the same colors from your monitor like I got from mine, because it's differently calibrated, and also, LCD or CRT isn't the same for a digicam (mine is CRT). Important is that you get similar color as from a "reference camera" from the same monitor (or you can print the image and take a photo of it)