500d development - page 210 - DSLR Hack development - CHDK Forum
« previous next »
Pages: 1 ... 205 206 207 208 209 [210] 211 212 213 214 215 ... 249   Go Down

500d development

  • 2487 Replies
  • 901259 Views
*

Offline pzotalis

  • *
  • 3
Re: 500d development
« Reply #2090 on: 13 / December / 2011, 15:08:49 »
Advertisements
Hello,

I'm new in ML. I have a 500D and still testing this great future you are developing. Sorry my english, I'm from Argentina.

I have a suggestion to make regarding the ISO options in Photo mode. Sometimes I think it would be great to be able to set a limited Auto ISO by establishing minimun and maximun ISO.
For instance, I believe that 60D has Adjustable Auto ISO limit.

For example, setting 400 as min and 1600 as max.

It would be a usefull tool.

Thanks for all the work you are doing, the 500D owners really appreciate!!
Logged
*

Offline mk11174

  • ****
  • 417
Re: 500d development
« Reply #2091 on: 13 / December / 2011, 16:02:11 »
Quote from: Gary on 13 / December / 2011, 12:42:42
Please post your autoexec.bin for testing.  Also the 422 python script.  Maybe I can fix the overlapping.  At what point will Alex include this in the official build?  Before or after we fix the overlapping?
Not sure if Alex will include this in his updates, I am sure if he feels its nessacery he will. Here are both the edited 422-jpg.py script and the autoexec.bin my bin is everything from Alex latest update except the addition of what I posted plus for me, i dont have any use for lcd brighness and volume control and since our flash button is useless with ML I have the iso and kelvin set for sensor shortcut, hope thats ok.

http://www.mediafire.com/file/3fq43odeoho7j89/autoexec.bin
« Last Edit: 14 / December / 2011, 16:28:00 by mk11174 »
Logged
*

Offline Gary

  • *
  • 43
Re: 500d development
« Reply #2092 on: 14 / December / 2011, 15:59:24 »
I think you sent me the wrong python script?
or the wrong autoexec?
They do not match.
What mode should I use?
P mode on dial then Liveview or Video mode idle?
The resulting filesize is the same for both.
Maybe it is user error and I need some instructions.

The size of my file is 29,830,400
What should it be?
Line 138 is elif len(data) % 1024*680*2 == 0:
There is only 1 call to convert_422_hires(input,output)
Unfortunately 29,830,400 mod 1024*680*2 not = 0 so it raises an exception in line 141:
Unknown image size.

I should change to 944,632 in how many places?
« Last Edit: 14 / December / 2011, 16:05:30 by Gary »
Logged
*

Offline mk11174

  • ****
  • 417
Re: 500d development
« Reply #2093 on: 14 / December / 2011, 16:23:23 »
Quote from: Gary on 14 / December / 2011, 15:59:24
I think you sent me the wrong python script?
or the wrong autoexec?
They do not match.
What mode should I use?
P mode on dial then Liveview or Video mode idle?
The resulting filesize is the same for both.
Maybe it is user error and I need some instructions.

The size of my file is 29,830,400
What should it be?
Line 138 is elif len(data) % 1024*680*2 == 0:
There is only 1 call to convert_422_hires(input,output)
Unfortunately 29,830,400 mod 1024*680*2 not = 0 so it raises an exception in line 141:
Unknown image size.

I should change to 944,632 in how many places?
hmm, ur right it is the wrong script, i reuploaded it right after i sent the first link, i guess i sent it to fast cause it kept the wrong one, ill re up it, the autoexec is the right one though just got to resend script, sorry.

Here is the correct 422-jpg script : http://www.mediafire.com/file/xoougnyck2f9x3x/422-jpg.py
« Last Edit: 14 / December / 2011, 16:27:22 by mk11174 »
Logged
*

Offline mk11174

  • ****
  • 417
Re: 500d development
« Reply #2094 on: 14 / December / 2011, 18:24:57 »
As far as I can see, the overlaping doesnt seem to be in the script, cause if u use the script to separate all the section by a few pixels you can see that theres no way they will ever line up, it looks more like a scan line issue that is captured during most of them, funny, 3x4 and 3x3 seem perfect so far, no scan line caught yet, all other modes seem to catch one, the shift has nothing to to with the individual images its just usuall in 1 or 2 places across the screen just like scanlines, take 3x3 and 3x4 see if urs is fine every try like mine, just might be lucky no scan line happened during them though.

update, lol, was wrong they all have scan line issues, maybe im wrong, but they sure seem like scan lines to me, no matter how much you even try to correct the image, sometimes the scan line is only on one side so as you adjust the settings in the script all the section will adjust evenly so if you try to fix the bad scan line area the rest of the image will get messed up.
« Last Edit: 14 / December / 2011, 18:37:29 by mk11174 »
Logged
*

Offline mk11174

  • ****
  • 417
Re: 500d development
« Reply #2095 on: 14 / December / 2011, 19:56:56 »
Hey Alex, just wondering if u can help me understand something, to see if i got it right in my head??? Is the reason we can not force a record at 5x zoom in movie mode because we can not tell the cam to record a movie on a differ buffer and in movie mode the cam is recording to its default movie record buffer even if we force zoom, and the only thing we can do is capture the buffer and do a print out which is to slow obviously for an actual movie, is this how things work on the cam and with ML? Hope u can answer, I am just trying to understand how this all works.
Logged
*

Offline Gary

  • *
  • 43
Re: 500d development
« Reply #2096 on: 14 / December / 2011, 20:01:46 »
Thanks!  I think I understand what's going on, give me a bit of time to fix it.
Logged
*

Offline a1ex

  • *****
  • 671
  • ML dev
Re: 500d development
« Reply #2097 on: 15 / December / 2011, 01:06:17 »
For 5x recording, I think it will work if you disable sound. The FPS hack requires the same thing (otherwise goes to buffer overflow quickly).

The video can be "deinterlaced", but I'm sure there will be heavy compression artifacts.
Logged
*

Offline mk11174

  • ****
  • 417
Re: 500d development
« Reply #2098 on: 15 / December / 2011, 01:27:31 »
Quote from: a1ex on 15 / December / 2011, 01:06:17
For 5x recording, I think it will work if you disable sound. The FPS hack requires the same thing (otherwise goes to buffer overflow quickly).

The video can be "deinterlaced", but I'm sure there will be heavy compression artifacts.
Yeah, I already tried it, I didnt have any issues with buffer cause I already had audio turned off, but like you explain in a topic awhile back, the pitch is off so it records a scrambled video, is there a way to set a new pitch then during that mode or is it impossible because its recording to the memory card through camera not ML, is it unadjustable other then in something like silent picture type captures. My code is all adjusted right now for ML silent pictures, I can capture good 5x stills fine, but even though the video looks fine while recording at a forced 5x mode, it still plays back scrambled.
Logged
*

Offline a1ex

  • *****
  • 671
  • ML dev
Re: 500d development
« Reply #2099 on: 15 / December / 2011, 02:18:16 »
The general idea to unscramble it is to load every frame (maybe in 422-jpg), convert it to plain RGB data and then encode it with a different pitch.

In memory, they are stored like this:

Code: [Select]
[ a b c
  d e f
  g h i ] => [a b c d e f g h i]

That would be with a pitch of 3. If we change it to 4, we'll get:
Code: [Select]
a b c d
e f g h

I guess this needs to be done in RGB colorspace. But I don't think the results will be nice, especially due to 420 sampling which happens after scrambling. The compressor will "see" a lot of false detail, which will force a low quality factor.
Logged
Pages: 1 ... 205 206 207 208 209 [210] 211 212 213 214 215 ... 249   Go Up
« previous next »
 

Related Topics