Canon 1100D / T3

Started by a1ex, June 14, 2012, 04:50:54 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

a1ex

Quote from: Danne on January 31, 2018, 07:59:16 PM
And files comes out clean and nice without errors. But I don´t get errors with today´s raw_video_10bit_12bit either even without the fix.

Well, it's easier when Canon code does the memory allocation for us (that was the point of these last tests - to use their buffer, like we do on all other models). This camera has the lowest amount of RAM out of all other ML-enabled cameras, so it's best if we don't have to allocate large stuff just to get raw LiveView data. However, should we ever need to do that (maybe if anyone would like to implement full-res LiveView for timelapse), the raw backend is now ready for that (cc @ these folks).

Quote from: Danne on January 31, 2018, 07:59:16 PM
Anybody else tired of rescale and convert PPM files one by one by the way? ffmpeg does the trick:
for i in *.PPM ; do
    ffmpeg -i "$i" -pix_fmt rgb24 -vf scale=500:-1 $(basename "${i/.PPM}").tif
done


What about this?

for i in *.PPM; do
    convert $i -resize "500x500>" ${i%.PPM}.png
done


(for pixel peeping, like I did to estimate the raw buffer size, I actually prefer the screenshot non-resized)

dfort

Hey Danne - nice work on the 1100D. It would be great to get some simple silent DNG files from the various settings and run the raw_diag module so we could check that the skip values are correct. Looks like the zoom mode needs some adjustment.

Danne

convert is nice too :).

@dfort
A little dinner break and I see if I get the energy back today or I´ll just continue in the morning...

a1ex

Quote from: Danne on January 31, 2018, 08:51:18 AM
malloc error :((sad smiley from liveview window)

Solved. Took many hours to rework the memory backend (found lots of other issues besides this one), but now it runs on all models in QEMU. Overall, the memory backend should be a lot more robust.

Torture test: old-style Lua (without umm_malloc) on 1100D (the camera with lowest amount of memory). It finally worked! 8)

PR open.

Danne

Epic progress. No cams here but will test as soon as possible.

a1ex

Also looking for a high-res photo of the camera display (large enough to see the individual pixels), showing ML menu with any recent build.

If it's hard to get the entire screen in focus, just focus on some problem areas (e.g. some icon or fine print that doesn't look well).

Floont

Quote from: a1ex on February 27, 2018, 06:20:45 AM
Also looking for a high-res photo of the camera display (large enough to see the individual pixels), showing ML menu with any recent build.

If it's hard to get the entire screen in focus, just focus on some problem areas (e.g. some icon or fine print that doesn't look well).

I've been using magiclantern-Nightly.2017Aug07.1100D105, but per a1ex's request, I loaded magiclantern-lua_fix.2018Feb24.1100D105 and took the two attached photos.  (This version is effectively non-functional for me - it displays a large number of error messages at launch, most of the menus are just black pages with an icon at the top, and I can't navigate within the menus themselves, but that's not what the question was about (so I'll return to  magiclantern-Nightly.2017Aug07.1100D105 for now)).
Hopefully these  images are close to what you're looking for.  I can try other things if it's helpful.
(Please note that I have a scratch-guard coating on my display, so there might be some artifacts from that - if so, such artifacts should be consistent between the two photos.)








a1ex

Ha, the screen looks very similar to the one from VxWorks models.

Press MENU to get out of that screen (you have it in the regular builds too).

Can you try a few more different screens? In particular, raw histogram has some fine details (the vertical bars). A large console should also help. Also, a large zoom on fine print, please (even if it's just a small part of the screen in the picture).

Vertically, things are pretty clear - odd lines are missing (first line being 0). Horizontally, I'm not very sure what happens. Best guess: all columns end up used, half of them on odd physical lines, and the other half on even ones.

Floont

Quote from: a1ex on February 27, 2018, 03:37:29 PM
Can you try a few more different screens? In particular, raw histogram has some fine details (the vertical bars). A large console should also help. Also, a large zoom on fine print, please (even if it's just a small part of the screen in the picture).
As I said, this new version doesn't play well with my camera.  I can't get the histogram to display - menu item is on, but grayed out.  It works in the 2017Aug07 version, if that's helpful (not sure if you're looking for something new or just how the 1100D display looks in general).

Fine print from magiclantern-lua_fix.2018Feb24.1100D105 version:





Histogram from magiclantern-Nightly.2017Aug07.1100D105 version:



a1ex

Yes, that helps. Can you (or anyone else who might have an extreme macro lens) zoom on the histogram even more? I'd like to find out how the subpixels are arranged and how the display buffer is interpreted by the hardware (just as a curiosity, not something of high priority).

BTW, you have some modules from the older version, that's the reason you've got the errors.

Whenever something is grayed out, it also has a message at the bottom of the screen. Can you read it?

Floont

The four photos I posted were taken with my iPhone SE.  I dug out my old A590 and S30, both of which have macro features, but neither one got an image sharper or higher-resolution than the iPhone pic I already posted.  Hopefully, someone with a better second camera/lens will chime in.

I need to leave for a while, will look at the 2018Feb24 version again another time - I'll try a completely clean install to see if it clears up any of the issues I'm seeing.

fitopy

I would like to know if there will be firmware for the t3 1100d with firmware 1.0.6 because the one that currently exists is for firmware 1.0.5, thanks :)

dfort

Doing a minor firmware update like 1.0.5 -> 1.0.6 on the 1100D should be relatively easy.

https://www.magiclantern.fm/forum/index.php?topic=19417.0

It requires a firmware dump and some time. I can help but someone needs to get this started. You can use the portable ROM dumper that is on the bottom of the Nightly downloads page. Don't post a direct link to the firmware dump! Just post when you have it and we'll take it from there.

Instructions on how to run the portable dumper:

https://www.magiclantern.fm/forum/index.php?topic=16534.0

dfort

Thanks @Danne - Another Canon firmware update ML port in progress:


Danne

Cool! Glad to be able to help.

dfort

It's Alive!





That was the easy part, several modules aren't compiling but that's happening on the unified 1100D.105 nightly builds too.

Getting some interesting message in QEMU:

  1583:  7037.440 [MC] cam event guimode comp. 1
  1623:  8292.096 [STARTUP] too heavy... free : 442KB
  1624: 14325.504 [STARTUP] too heavy... free : 442KB
  1625: 20293.376 [STARTUP] too heavy... free : 442KB
  1626: 26326.784 [STARTUP] too heavy... free : 442KB
  1627: 32294.656 [STARTUP] too heavy... free : 442KB
  1628: 38328.064 [STARTUP] too heavy... free : 442KB
  1629: 44295.936 [STARTUP] too heavy... free : 442KB
  1630: 50329.344 [STARTUP] too heavy... free : 442KB
  1631: 56297.216 [STARTUP] too heavy... free : 442KB
  1632: 62330.624 [STARTUP] too heavy... free : 442KB
  1633:  2762.496 [STARTUP] too heavy... free : 442KB
  1634:  8795.904 [STARTUP] too heavy... free : 442KB
  1635: 14763.776 [STARTUP] too heavy... free : 442KB


I'll post a test build on my downloads page but the selftest module isn't working so it will be just to see if it is working the same as the 1.0.5 nightly version. Seems like it would be better to test this camera on the raw_video_10bit_12bit branch because more modules are working on that branch.

One question, is anyone actually using this camera and is willing to run it through some tests?

DavidSamish

This was my first DSLR, now it's my second backup camera. I use ML on it and am interested in trying out the latest version. The other two Canons I have are a 70D and 5DMkIII.

dfort

@DavidSamish - You can start by trying the 1100D.106 test build I posted on my downloads page and reporting your findings. If it works and seems about as stable as the nightly build I can prepare builds using the various branches so we can run some tests on real hardware.

aplinhares

Thanks for the update, @dfort !!!
will test too the 1.0.6 build

dfort

@aplinhares - thanks. There's a 1100D Canon Firmware for Testers.zip package on my downloads page with both the Canon 1.0.5 and 1.0.6 firmware updaters so you don't need to search for them to go back and forth. If the test build works about the same as the current nightly we can start testing some other branches that give the 1100D some more functionality.

DavidSamish

@Dfort: I upgraded my firmware to Canon's 1.0.6 alright, but with a different card loaded with your 1.0.6 ML files; ML folder and autoexe.bin file, I can't seem to do the ML firmware update from the Canon-menu-firmware route. Do I need the ML-SETUP.FIR file from the other card? Nope- that didn't work. What am I doing wrong?

dfort

@DavidSamish - This is an early test build and doesn't have an ML-SETUP.FIR file for the 1.0.6 firmware yet. If you didn't uninstall ML before doing the Canon firmware update then your camera boot flag is still set so put the 1100D.106 ML files on a bootable card -- a card that has had ML installed should have the bootflag set. You can also set the bootflag on your card using either EOScard for the PC or MacBoot for the Macintosh. If all else fails, go back to 1.0.5, install 1100D.105 and make sure it is working then update again to 1.0.6 making sure you don't uninstall ML or format the card in a way that erases the bootflag.

Once I get some feedback that the test build actually works on the camera I'll ask for a ML-SETUP.FIR for the 1100D.106.

a1ex

One thing I'd like to be double-checked: the Q button code. Debug -> Show GUI events -> press Q and write down the event code.

On 1200D, this button code appears to be different than what we have assumed.

dfort

The 1100D was falling behind other cameras on the unified branch so I brought it up to date with this pull request. New test builds for the unified branch for both 1.0.5 and 1.0.6 posted on my downloads page. Once these check out we can move on to the lua_fix tests.

Also, please check the previous post about the Q button code.

DavidSamish

@dfort: I got the 1.0.6 firmware and the ML 1100D106 to work together. I had to degrade to 1.0.5 and re-upgrade, as you mentioned. The MacBoot didn't work; I got "The Java class file "MacBoot.class" could not be launched."

I am getting an error message: 'Error loading "ML/MODULES/t3_106.sym" File does not exist." and on the bottom of the screen "missing symbol file: ML/MODULES/t3_1006.sym : File does not exist" and
"save configs...
save configs..."

So far so good...