CHDK 1.4 release + bug reports

ursamajor

64
SX150IS, SX510HS

Re: CHDK 1.4 release + bug reports

« Reply #40 on: 24 / January / 2017, 12:29:18 »

Advertisements

Aaa, sorry then, and thanks!

Logged

SX510HS, SX150IS (retired)

srsa_4c

4451

Re: CHDK 1.4 release + bug reports

« Reply #41 on: 21 / April / 2018, 15:12:50 »

I wasn't sure where to post this, so I'm posting it here as it probably qualifies as a bug.

There is an issue affecting cameras that can record video in non-video modes.
When recording multiple videos without switching the camera off, the CHDK bitrate / remaining time display gets more and more incorrect with each recording.

I believe the logic inside gui_osd_draw_movie_time_left() (core/gui_osd.c) fails for recordings made in modes other than video.

The following patch seems to help, although I'm not sure it's optimal and without side effects.

Code: [Select]

Index: core/gui_osd.c
===================================================================
--- core/gui_osd.c	(revision 5010)
+++ core/gui_osd.c	(working copy)
@@ -618,6 +618,14 @@
             }
         }
     }
+#ifdef CAM_HAS_VIDEO_BUTTON
+    else {
+        // seems to be needed for cameras that can record video in non-video modes (dedicated movie button)
+        // without this workaround, bitrate / remaining time calculation will be incorrect for subsequent recordings
+        // todo: can this cause any side effects?
+        init = 0;
+    }
+#endif
 }
 
 void gui_osd_draw_ev(int is_osd_edit)

I first noticed this while working on the D6 movie support code, but managed to reproduce it on my Ixus115.
Can somebody with a dedicated movie button camera confirm the issue?

Logged

blackhole

937
A590IS 101b

Re: CHDK 1.4 release + bug reports

« Reply #42 on: 22 / April / 2018, 10:39:28 »

Quote from: srsa_4c on 21 / April / 2018, 15:12:50

I wasn't sure where to post this, so I'm posting it here as it probably qualifies as a bug.

There is an issue affecting cameras that can record video in non-video modes.
When recording multiple videos without switching the camera off, the CHDK bitrate / remaining time display gets more and more incorrect with each recording.

I believe the logic inside gui_osd_draw_movie_time_left() (core/gui_osd.c) fails for recordings made in modes other than video.

The following patch seems to help, although I'm not sure it's optimal and without side effects.

Code: [Select]
Index: core/gui_osd.c =================================================================== --- core/gui_osd.c (revision 5010) +++ core/gui_osd.c (working copy) @@ -618,6 +618,14 @@ } } } +#ifdef CAM_HAS_VIDEO_BUTTON + else { + // seems to be needed for cameras that can record video in non-video modes (dedicated movie button) + // without this workaround, bitrate / remaining time calculation will be incorrect for subsequent recordings + // todo: can this cause any side effects? + init = 0; + } +#endif } void gui_osd_draw_ev(int is_osd_edit)
I first noticed this while working on the D6 movie support code, but managed to reproduce it on my Ixus115.
Can somebody with a dedicated movie button camera confirm the issue?

I see something similar to a sx410 camera.
The remaining time after a dozen recorded clips starts to grow instead to falling.
Much stranger things happen if I delete the last recorded clip.
After deleting of the last recorded video,for the remaining time and data rate are displayed completely wrong data.
Regardless of the incorrect displayed data rate (KB/s), the size of the recorded file corresponds to the settings for the desired bitrate.
With the restart of the camera everything returns to normal.

Logged

http://astrofoto.pondi.hr

reyalp

14080

Re: CHDK 1.4 release + bug reports

« Reply #43 on: 23 / April / 2018, 02:06:22 »

Quote from: srsa_4c on 21 / April / 2018, 15:12:50

I wasn't sure where to post this, so I'm posting it here as it probably qualifies as a bug.

There is an issue affecting cameras that can record video in non-video modes.
When recording multiple videos without switching the camera off, the CHDK bitrate / remaining time display gets more and more incorrect with each recording.

Can somebody with a dedicated movie button camera confirm the issue?

Good find. I confirmed this on elph130, and the patch seems to resolve it.

I don't see any obvious problems, but the but moving the record_running logic before the main if might make it easier to follow, as in the attached.

reyalp_vid-time-1.patch (1.42 kB - downloaded 35 times.)

Logged

Don't forget what the H stands for.

srsa_4c

4451

Re: CHDK 1.4 release + bug reports

« Reply #44 on: 24 / April / 2018, 14:15:25 »

Thanks for both of you. Meanwhile, I could confirm the existence of this issue on an old S3 IS too.
@reyalp
I'm OK with your patch, please check it in (both trunk and 1.4, I guess).

Logged

reyalp

14080

Re: CHDK 1.4 release + bug reports

« Reply #45 on: 25 / April / 2018, 00:25:40 »

Quote from: srsa_4c on 24 / April / 2018, 14:15:25

Thanks for both of you. Meanwhile, I could confirm the existence of this issue on an old S3 IS too.
@reyalp
I'm OK with your patch, please check it in (both trunk and 1.4, I guess).

Checked in, trunk r5011, 1.4 r5012

Logged

Don't forget what the H stands for.

Caefix

945
Sorry, busy deleting test shots...

Re: CHDK 1.4 release + bug reports

« Reply #46 on: 15 / October / 2019, 12:17:09 »

Some kind of millennium bug?

shooting.c line 41
typedef struct {
short av96;
short tv96;
short sv96;
short subj_dist; // should´t such values be int for newer /hizoom cams?
short nd_filter;
} PHOTO_PARAM;

Nice?
D:\hostluaPortable\CHDK-Shell\trunk\trunk-5279\platform\generic\wrappers.c
line 263 : _MoveFocusLensToDistance((short*)&newpos);
lolevel.h line 22
extern void _MoveFocusLensToDistance(short *dist);

« Last Edit: 15 / October / 2019, 13:43:47 by Caefix »

Logged

All lifetime is a loan from eternity.

reyalp

14080

Re: CHDK 1.4 release + bug reports

« Reply #47 on: 15 / October / 2019, 12:50:09 »

Quote from: Caefix on 15 / October / 2019, 12:17:09

Some kind of millennium bug?

shooting.c line 41
typedef struct {
short av96;
short tv96;
short sv96;
short subj_dist; // should´t such values be int for newer /hizoom cams?
short nd_filter;
} PHOTO_PARAM;

Thanks. I checked in a fix as mentioned in https://chdk.setepontos.com/index.php?topic=13875.msg141593#msg141593

Quote

D:\hostluaPortable\CHDK-Shell\trunk\trunk-5279\platform\generic\wrappers.c
line 263 : _MoveFocusLensToDistance((short*)&newpos);

This should not be a problem. Some firmware _MoveFocusLensToDistance may expect a short, others an int, but the pointer will work for either.

Logged

Don't forget what the H stands for.

Caefix

945
Sorry, busy deleting test shots...

Re: CHDK 1.4 release + bug reports

« Reply #48 on: 28 / April / 2020, 12:02:50 »

Test/filetest.lua now prints always something.

Code: [Select]

--[[
@title file browser test
@chdk_version 1.3
]]

function restore()
play_sound(0)
print(f or "not f")
end

f = file_browser("A/CHDK/SCRIPTS") or "<Folder!>"
play_sound(5)
print(f)

Logged

All lifetime is a loan from eternity.

Caefix

945
Sorry, busy deleting test shots...

Re: CHDK 1.4 release + bug reports

« Reply #49 on: 12 / May / 2020, 14:43:18 »

Found my infrared Ixus300 with #5163

Something happened on the road to #5285 (#5229?), the oldest versions I archived.
Using pure #5163

Code: [Select]

--[[
@title file browser test
@chdk_version 1.3
]]

f = file_browser("A/CHDK/SCRIPTS")

print(f)

returns the expected nil, the later breaks script if folder chosen with "menu"

Edit2: Suspicious zone:

fselect.flt[5205] repairs #5229.
Edit: Not missing:

IR-look.JPG (1939.41 kB, 3648x2736 - viewed 62 times.)
Model:Canon IXUS 300 HS Exposure Time:1/80 seconds F Number:F/2.8 ISO Speed:125 Date Taken:2018:09:22 16:41:34 Metering Mode:Center Weighted Average Flash Mode:Flash did not fire; compulsory flash mode Focal Length:4.9 mm Color Space:sRGB

« Last Edit: 13 / May / 2020, 12:31:10 by Caefix »

Logged

All lifetime is a loan from eternity.