Okay thanks for the hints. I will continue testing.One thing I'm not quite sure about is: How do I set the script to continuous mode or to single shot mode? Do I have to set the interval time to 0, so that it immediately takes a new picture for continuous mode and otherwise it is in single shot mode?
Quote from: philmoz on 12 / September / 2013, 05:09:45Looks to me like the shot_histogram calculation and shot_histogram_delay are preventing CHDK from saving the DNG file before the firmware starts the process of conversion to JPEG.That could be it, but there seems to be something different in the 101a build with exposures over 1 second in continuous mode. The shot meters aren't garbage, like a wrong raw buffer address. All 4 meters just get brighter, but their relative brightness rank to each other stays the same, like they're still reading the same picture. There are 4 meters, each measuring a different area of the picture. The brightness difference for <1 second shots and >1 second seems to be constant for an individual meter. Meter readings are in ev96. They all get brighter, but not by adding a constant ev96 value. It's looks more like the ev96 meter values are MULTIPLIED by a constant.
Looks to me like the shot_histogram calculation and shot_histogram_delay are preventing CHDK from saving the DNG file before the firmware starts the process of conversion to JPEG.
The purple tinted DNG images look like the JPEG tone curve has been applied to the RAW data. This would account for the colour tint because red and blue pixels have a different tone curve to green pixels.
Quote from: philmoz on 12 / September / 2013, 16:02:05The purple tinted DNG images look like the JPEG tone curve has been applied to the RAW data. This would account for the colour tint because red and blue pixels have a different tone curve to green pixels.That would explain the changes in meter readings. You're right that a sensor gain change would probably increase all the pixels by the same, linear amount.I decided to test DNG file saving on my own G1X_100g in continuous mode with exposure over and under 1 second. I was surprised to see that the DNG files with 1.3 second exposure were all distorted in color, while the files with 0.8 second exposure were all normal. All my camera settings are off.It sounds like the meter readings are correct on my G1X because I'm done computing them before the camera starts changing the raw buffer. But saving DNG files takes a lot longer, so they're always color distorted. The 101a update must start processing the raw buffer sooner for >1 second continuous mode exposures.Phil, will you take some DNG pictures with exposure 1.3 seconds, Tv mode, continuous drive, and see if it happens on your G1X too? I hope so, because you're the one who can figure it out! Thanks.Here's a screenshot from Lightroom of the 0.8 and 1.3 second DNG pictures (attached):[EDIT] The color distorted DNG files were all taken at 400 ISO. When I tried it at 100 ISO, the 1.3 second DNG's were normal.The DNG files taken by my time lapse script (Av continuous mode) were all normal, including setting the camera to 100 or 400 ISO before starting the script. The script set the exposure to 2 seconds 400 ISO, and the DNG's were normal. Setting the camera to 400 ISO (Tv 2 second continuous mode) produced all color distorted DNG.I noticed you changed the way you find the raw buffer address recently to a table lookup instead of 2 fixed addresses. Could that be related? Maybe there's a another buffer where the raw data is processed into the jpg. The raw save routine could be picking up the address of that buffer by mistake, depending on the timing. It might be useful to display the raw buffer address (static char *rawadr) in core_spytask.
I can reproduce the problem on my G1X (1.00f).Whatever is happening it appears to be in another task (not capt_seq_task).I don't have time to investigate any further - I would suggest using the built in RAW support instead of CHDK DNG in this case.
But it's interesting: I've always shot time lapses in single shot mode. Now I tested in the continuous mode and it works perfectly.
New user here, excited to try ptp with multiple cameras.I tested two G1x (101a and 100e) with usb switch in sync mode and found it was not very reliable, crashing fairly often.The romlog file is attached.I am surprised that no one has reported this problem.Do you know if there is any way of fixing it ?
1) What CHDK settings remote settings are being used
2) When does the crash occur? e.g. when you start the shot, just randomly, when the picture is saving...
3) Does it happen on both cameras, or just one?
4) What canon shooting mode and settings are you using (e.g. P, M, Auto), manual focus, auto focus etc.
5) Are you using both PTP and the USB remote?
You might try turning of Canon features like face detect and continuous autofocus.
Quote1) What CHDK settings remote settings are being usedEnable remote = YesSwitch type = OnePushControl Mode = NormalEnable synch = YesSynch delay = NO, 100
While switch is pressed,screen blanked and camera waiting for switch release.Quote3) Does it happen on both cameras, or just one?Both, except it does not happen if a camera is upside down !
Seems to be the same problem, maybe zeroing the face-detect magic number will fix ?
Started by technojit Feature Requests
Started by x2lazyx General Help and Assistance on using CHDK stable releases
Started by eXistenZ CHDK Releases
Started by Metalzone CHDK Releases
Started by ironsting Feature Requests
