Tragic Lantern for EOS M

Started by coutts, April 17, 2013, 01:43:28 AM

Previous topic - Next topic

0 Members and 2 Guests are viewing this topic.

1%

I did unsqueeze of 5:3 so it would be multiplied by 3/5... I'll try 16:15 next I guess. I'll go back to af_raw if none of the raw types remove the dots.

mixer2

Try 1/1, i think 16/15 is a measurement inaccuracy... I just took a video of a circle and measured, how much it's squeezed. It's not a very accurate way of getting the squeeze factor. And 16/15 is so near to 16/16 or 1/1, that i can't imagine, that the difference doesn't come from inaccuracy because of the measurement.

coutts

Quote from: qsara on June 10, 2013, 02:42:42 PM
to use on EOS M? or just kidding?
hope he knows eos m uses sd not cf

mixer2

read the posts above... he wondered why his 5d3 wasn't able to record continuously in high resolution, while using sd-card. it was just OT.

1%

1:1 is no unsqueze at all, just like default.

So no raw type helps.. dots come on. AF_RAW gets rid of dots but causes the "peaking" effect.

mixer2

exactly... i think the unsqueezing was the reason why it was squeezed. The current factor is 5/3. if i take a frame (1280x688) and undo the 5/3 squeezing i get 688*5/3 ~= 1146. If i resize it to 1280x1146, it looks just perfectly unsqueezed.

qsara

Quote from: mixer2 on June 10, 2013, 07:22:52 PM
read the posts above... he wondered why his 5d3 wasn't able to record continuously in high resolution, while using sd-card. it was just OT.

I don't have a 5D.III troll. read my signature.

this is an EOS M thread and I'm talking about it. you're the one writing about your 5D.III. go that thread and write you've ordered whatever for your camera.

mixer2

You quoted "Canon eos m" and he has. I just gave him an answer to his question, even if it was OT...

Quote from: Canon eos m on June 10, 2013, 10:16:09 AM
These recent days, I figured out a way to make my 5D Mark 3 RAW capable using the ML upgrade. I tried taking a few videos with it using a Class 10 95MB/s Scandisk SD Card. I now manage to capture about 100 frames before the buffer runs out. Don't know why and don't know what it means not to be able to do more .... but .... the fact that I recorded RAW has got me loving it and wanting more.

qsara

I've quoted %1 and asked about modules, read the post, again, and again, then again.

mixer2

Quote from: qsara on June 10, 2013, 02:42:42 PM
Quote from: Canon eos m on June 10, 2013, 01:20:31 PM
Ordered the Komputerbay CF 128GB 1000x. Should be delivered any day now.
to use on EOS M? or just kidding?

Are you kidding now?

1%

Original was not unsqueezed and it still had to be resized to be fixed.

Some new hope in the dead pixel remover that just got committed. Next one I guess I'll just do no unsqueeze and fix it in post.

mixer2

hurm, then i really don't understand it at all.
what's in the source:
int correct_height = max_res_x * 3 / 5;

and we get a squeezed 1280x688 what looks perfectly unsqueezed when resized to 1280x1146.
correct_height = 688
max_res_x => 688/3*5 ~= 1147
if you use:
int correct_height = max_res_x;
correct_height should be 1147 or 1146, depends how it's rounded.
and 1280x1146 (or 1280x1147) should be the result, which should look perfectly unsqueezed. what exactly is wrong in my calculation?

1%

Will try it and see what  happens.

mixer2

I hope it will work. As said before, i don't know anything about how all the stuff works... just playing around with the numbers and guess a solution :D

a1ex

Shouldn't be int correct_height = max_res_x * 2 / 3 ?

The image covers usually a 3:2 or a 16:9 area. I don't see why Canon would have designed it for 5:3... and from the dng posted a while ago, with the square floor pattern, it should be 3:2.

mixer2

The image was 1152x656, but it was already unsqueezed by factor 16/9.
To undo that 656*16/9 = 1166
Then do the 3/2 unsqueeze:
1166*2/3 = 777

But if you resize it to 1152x777 it still looks squeezed.

What's a bit strange is, that if i use the undone 1166 height, it's minimal vertically streched. But, if i do the undo calculation with 5/3:
656*5/3 = 1093 and i resize the image to 1152x1093 and it's perfectly unsqueezed. So 1:1 might be not perfect.
So i'm back at the:
int correct_height = max_res_x * 15 / 16;

still just guessing, sorry...

a1ex

It's 1728x672, resize to 3:2 (1728x1152) and it looks OK to me.


1%


mixer2

hurm... that does look correct... but how have you exactly calculated the 1728x1152?
672*16/9*2/3 = 796

don't know what i do wrong... but... if you know the solution everything is good xD

is one of you in the irc, to discuss that more directly than in the forum. would love to understand what i did wrong.
Ah, i see. you calculated it with the width:
1728/3*2.... but... when the source, at the time the frame was taken, was:
int correct_height = max_res_x * 9 / 16;
why was the resulting vertical resolution 672???
1728*9/16 = 972

If it was 5/3 that time it wouldn't result in 672 too.

mixer2

Okey, tested the new version.
2 things are annoying:
1. have to disable and reenable "Use photo mode", to enable raw recording
2. the guessing mb/s guess in the menu is completely different from the one mb/s it shows while recording (22mb/s vs ~37mb/s). don't know which one is correct.

The dngs we get are still squeezed, but it tells in the menu that it has to be unsqueezed in post (with the correct ratio). Is that how it should be. Will the unsqueezing never be done automatically and has to be done in post all the time?

1%

Heh, set it on by default.. .will have to fix it so it actually switches to fake movie mode.

Escaperoute

Hi,
I just can't seem to uninstall ML on my camera completely, i follow all the instruction completely.

I have some experience with ML, it run smoothly on my 550D but on my EOS-M 18-55 mm shutter release bug is really annoying.
When i install the ML on my camera. it run smoothly, but when i enter the Menu setting i can't seem to find firmware update option with ML signature like on my 550D, when i update with ML memory card installed it just install ML again, Maybe bootflag cannot be remove.

22mm lens work fine. But i use 18-55 mm mostly, can anyone help me. i've read all the post but nothing seem to help me. I'm freaking out again since my 18-55 mm lens work sometime and sometime it just don't with or without ML on SD card.

qsara

Quote from: Escaperoute on June 10, 2013, 10:12:21 PM
Hi,
I just can't seem to uninstall ML on my camera completely, i follow all the instruction completely.

I have some experience with ML, it run smoothly on my 550D but on my EOS-M 18-55 mm shutter release bug is really annoying.
When i install the ML on my camera. it run smoothly, but when i enter the Menu setting i can't seem to find firmware update option with ML signature like on my 550D, when i update with ML memory card installed it just install ML again, Maybe bootflag cannot be remove.

22mm lens work fine. But i use 18-55 mm mostly, can anyone help me. i've read all the post but nothing seem to help me. I'm freaking out again since my 18-55 mm lens work sometime and sometime it just don't with or without ML on SD card.

copy all the files on your card to your computer,
format the card (if exfat) to fat32, (if fat32) to exfat,
move all the files again to the card,
use EOSCard utility to make your card bootable for magic lantern,
insert the card to the camera, use firmware update option, when you see the green text just switch to movie or photo mode.

Escaperoute

Thanks for the fast reply,

But what i'm trying to do is to get the ML loaded on my camera with 18-55 mm lens working, i've successfully uninstalled it but i can't seem to get it work with 18-55 mm attached.
Any comment??

1%

The lens thing is a bug I think.. hence canon update... all of my normal lenses work with the adapter and so does the pancake... btw modules seem to load up in photo mode now.