Canon 100D / SL1

Started by nikfreak, October 19, 2015, 10:41:29 PM

Previous topic - Next topic

0 Members and 2 Guests are viewing this topic.

dfort

Quote from: IDA_ML on February 08, 2018, 06:35:48 PM
Am I missing something?

Yes.

The frame rate is tied to the video mode needed to get to that particular raw buffer. In this case it is the raw buffer that is used when the Canon menu is set to 1280/60.

OlRivrRat

   Just moved over to MacOS 10.11.6 to view the Results of My previous FRSP Tests & found that

the DNGs that were saved to disc by ML show as Solid Black when Viewed by Preview, where as the

DNGs that come out of the MLV when "Switched", View in Preview just fine.

        The "Un-Preview-able" DNGs do open just fine in RPP.
ORR~DeanB  ~~  80D-ML  &  SL1+ML  &  5D2+ML  &  5DC+ML  &  70D+ML(AliveAgain)

canneloni

So, just to clear things up on how I use the FRSP feature i recorded a video of me using it and provides some sample images. As one can see in the video i select 1/25th as shutter speed, but the camera still sets it to 1/3th when I take the photo. The exif data in the sample images also show this shutter speed.

If there is some kind of error on my side using the module, please let me know so I can provide proper feedback whether or not the module works as it should.

Video and images can be found here:

https://drive.google.com/open?id=12mgivX0btxZed2HrOzAI1RFVEqhp8sUU

Edit:

@ a1ex: I did some touchscreen tests and recorded them. I was not sure if the log file was still logging, so I might have stopped early with the tests, but hopefully it's stills usefull. Link
100D.100B ; Canon 18-55 STM ; Canon 50 1,8 II ; Canon 75-300 4,0 - 5,6 III ; Sigma 17-50 2,8

dfort

@OlRivrRat and @IDA_ML - you might be interested in what I just posted on the crop_rec on steroids: 3K, 4K, 1080p48, full-resolution LiveView topic about "Full-res LV" -- found another issue on the 700D.

That "Full-res LV" setting can be a bit confusing check out what the tool tip at the bottom says:



So is it "full resolution" or is it a "LV snapshot" ?? Maybe it changes depending if crop_rec is active or not?

IDA_ML

Thanks for noticing this, OlRivrRat and Dfort.  Obviously, it is a bug on the 100D but there are still much better ways of shooting time lapses that will also provide better results in terms of quality - no desqueezing (and quality loss) required in post.  Here is what I usually do:

1) In the Photo mode, with the Silent module, Intervalometer and Life View active.  This is the old classical "gradient" way that has been working for years.  It provides time lapse sequences in the full sensor resolution, also called FRSP.  With the Canon preview set to Hold, you can take a test frame to check focus, exposure and composition and make corrections if necessary.  When you are ready, you change the preview to Off and start the actual time lapse.  Caveat:  At certain intervalometer settings, the time lapse may look a little choppy since there is some dead time between frames. 

2) In the Movie mode and FPS override set to some small vallue, say 1 fps.  Metering and focusing in this mode is a pain - it takes too long the camera to respond, but once you have the settings right, the timelapse looks very smooth and quality is gorgeous!  Caveat: Maximum resolution on the 100D is limited to 2520x1080, (at 5x magnification).

These two ways require some practice and experimenting which is also valid for every ML function,  but once you get used to them, the results are really rewarding and well worth the effort.

dfort

Ok--so it looks like this works fine:

Quote from: OlRivrRat on February 07, 2018, 10:58:57 PM
   As Per suggestions from DFort I Reset the SL1 & Trashed the ML Settings Folder.

With only crop_rec, mlv_lite, mlv_rec & silent Modules Active

FRSP + Intervalometer produces >

      Setting                      Saved to Disc

FRSP FullRes   DNG             DNGs

FRSP FullRes   L-DNG           Lossless DNGs

FRSP FullRes   MLV               MLV

And this is invalid for the 100D:

Quote from: OlRivrRat on February 07, 2018, 10:58:57 PM
Canon Movie Res set to  1280x720/60FPS + ML Crop mode 3x3 720p

FRSP FullResLV   DNG            No Files

FRSP FullResLV   L-DNG         No Files

FRSP FullResLV   MLV            No File

       Build used >

magiclantern-crop_rec_4k.2018Feb05.100D101


Quote from: OlRivrRat on February 09, 2018, 05:38:04 PM
...Might be good to remove the Option from Cam's

on which it will not work.

Agreed. On the crop_rec module the cameras with the advanced features have different menus from the cameras with basic features.

https://bitbucket.org/hudson/magic-lantern/commits/58f3eab9038ad8745daa8627e1635cb402000886

a1ex

Can other 100D owners confirm issue #2846?

(possible interference with Canon audio controls)

OlRivrRat

                 @ Issue #2846

   SL1 + Rode StereoMic + magiclantern-crop_rec_4k.2018Feb05.100D101

On 1st attempt the Issue seemed to be there as whether RSM was On or Off I

was getting Audio in SL1. I then unplugged the RSM & Lost all Audio, replugged

RSM & still no audio. Noting that the Battery was Low I replaced the Batt' & now

All is working as it should >

RSM Plugged in & Off = No Audio

RSM Plugged in & On = Audio

   BTW ~ Still wishing that the Sound Menu was Not Hidden within the

Raw video Menus > Especially since Audio works whether Raw Vid is on or off.
ORR~DeanB  ~~  80D-ML  &  SL1+ML  &  5D2+ML  &  5DC+ML  &  70D+ML(AliveAgain)

a1ex

Quote from: OlRivrRat on February 11, 2018, 06:20:46 PM
   BTW ~ Still wishing that the Sound Menu was Not Hidden within the

Raw video Menus > Especially since Audio works whether Raw Vid is on or off.

Not sure what you mean by that, as the options from the raw video submenu only affect the sound operation when recording raw video. They should not interfere with H.264 - the settings for that are in Canon menu.

Furthermore, I don't see a real need to change MLV sound settings often. You load mlv_snd, you get sound in MLV without having to tweak anything else. You don't load it, you don't get sound. The audio meters should reflect that without any menu tweaking.

OlRivrRat

      @Alex

   Ooops ~ BODDBMe was mistakenly thinking that the Meters were On Only when

ML Sound was On.

   So let Me asked this > Do the Meters use up much Processing Power? Might it be

helpful in terms of Processor Burden to be able to turn them off if they aren't needed?
ORR~DeanB  ~~  80D-ML  &  SL1+ML  &  5D2+ML  &  5DC+ML  &  70D+ML(AliveAgain)

dfort

Just a wake up call for 100D users, this test hasn't been run yet so the 100D is using the dummy DEFAULT_RAW_BUFFER_SIZE in raw.c which is probably much smaller than this camera is capable of:

https://www.magiclantern.fm/forum/index.php?topic=5601.msg196632#msg196632

OlRivrRat

                     @DFort

       Was the test that Canneloni did on 04Feb18 insufficient?

6th Reply down from this >

https://www.magiclantern.fm/forum/index.php?topic=5601.msg196632#msg196632
ORR~DeanB  ~~  80D-ML  &  SL1+ML  &  5D2+ML  &  5DC+ML  &  70D+ML(AliveAgain)

dfort

Thanks for pointing that out but no, it looks like that test wasn't done right. Read the instructions here:

https://bitbucket.org/hudson/magic-lantern/commits/e799e89f67877825c6a4dc2868f87bed1196432b

If anyone wants to run this test but can't compile a build of the the allocate-raw-lv-buffer with CONFIG_MARK_UNUSED_MEMORY_AT_STARTUP, I'll compile a test build.

IDA_ML

Quote from: dfort on March 06, 2018, 08:59:49 PM
If anyone wants to run this test but can't compile a build of the the allocate-raw-lv-buffer with CONFIG_MARK_UNUSED_MEMORY_AT_STARTUP, I'll compile a test build.

Yes, do this please, Dfort, and give us detailed instructions on how to test.  Thanks.

dfort

Ok--turned out to be a little trickier than I thought. Made a special allocate_raw_lv_buffer build for the 100D and posted it on my downloads page.

Follow these instructions:

https://www.magiclantern.fm/forum/index.php?topic=5601.msg196632#msg196632

https://bitbucket.org/hudson/magic-lantern/commits/e799e89f67877825c6a4dc2868f87bed1196432b

OlRivrRat

       Have some Test Results > I Think > Where would You like them?
ORR~DeanB  ~~  80D-ML  &  SL1+ML  &  5D2+ML  &  5DC+ML  &  70D+ML(AliveAgain)


IDA_ML

Hey guys and happy 100D owners,

For those of you who may be interested in taking a look at how the 100D records RAW video at ISO 6400, here are a few examples:

https://we.tl/baHwIr5I18

The above clip has been saved in the NDxHR_HQ format, so if someone wishes to apply some noise reduction to it, just go ahead. It should play smoothly in your editor.

The conclusion in the end says that if shooting at such high ISOs, exposing to the right is absolutely mandatory. 

The above link will be active for 7 more days.

garry23

@IDA_ML

Ok, usual caveat from me, I'm not a videographer.

However, when shooting RAW, ETTR should only be used at 'low' ISOs, ie below, say, 1600ish, and camera dependent.

The reason is, that once you are in the ISO invariant zone, pushing the ISO further is a 'waste' of DR, ie signal or photons.

Bottom line: if shooting RAW, know your camera's ISO invariant zone and stay out of the pattern noise zone, but don't over push ISO, ie correct exposure in post.

IDA_ML

@Garry23,

What you are saying may be generally true but the examples in the above film show quite the opposite.  The 1-st and the 3-rd clips were shot with ETTR right before highlight clipping startted to occur.  This condition was achieved at F 4,0.  The 2-nd clip shows exactly the same scene but it was deliberately exposed one stop lower - at F 5,6. This clip is much noisier than the the other two shot at F 4,0.  This is to be expected since at F 5,6 less photons (50% less) reach the sensor meaning that the signal-to-noise ratio is lower, in fact 3 dB lower in terms of power.  You could repeat the same experiment and see for youself.

By the way, in your handheld bracketing script you use exactly the same phenomenon.  By setting the high ISO limit to ISO 6400 (and not to 1600 !!!) you blend with a photo that provides maximum signal-to-noise ratio in the shadows.  This is the reason why your resulting photo looks so amazingly clean in the dark areas.


garry23

@IDA_ML

You seem to be mixing aperture and ISO.

I'm only talking about ISO.

As I say, after about 1600-3200 you really are losing rather than winning.

The noise is not getting lower, but the signal is plummeting.

In my script I gave three max ISO options, 1600, 3200 and 6400. Your choice.

Bottom line: ISO invariance is real.


IDA_ML

Then how do you explain the fact that the second clip is much noisier than the other two, given the fact that all three clips were shot at ISO 6400 and the scene and lighting were exactly the same for all three clips?

I am not sure what you mean by saying: "after about 1600-3200 you really are losing rather than winning."  How would you expose a clip if it is so dark that the only way to get a proper exposure is to use high ISO?  Film at lower ISO and then push up the exposure in post?  Well, good luck with that!

garry23


a1ex

In regular video mode (with pixel binning, not with 1:1 crop), the sensor is a little "less invariant" at higher ISOs. On 5D3, for example, you should not think twice about using ISO 6400 in video mode (as the slope from 3200 to 6400 is far from "ISO invariant"), as opposed to photo mode and 1:1 crop, where the slope from 3200 to 6400 is a lot closer to "ISO invariant" or "diminishing returns" territory.

https://www.magiclantern.fm/forum/index.php?topic=10111.msg118553#msg118553
https://www.magiclantern.fm/forum/index.php?topic=19429.msg193332#msg193332

On small cameras (1.61x), the ISO invariant zone starts about 1.4 stops earlier (consider exposure equivalence - something I should implement on the ML UI), so the limit is probably 1600 or 3200. Above that, you will gain nothing in shadow SNR, but you will be (still) losing highlights.

http://www.clarkvision.com/articles/evaluation-canon-7d/index.html

garry23

@a1ex

Thanks for the insight.

As per my caveat. I don't do video, so I can only speak to the photo mode.

I haven't yet worked out the sweet spot for ISO invariance. Which is why I coded 1600, 3200 and 6400 as menu choices for max ISO in my hand holding script.

On my EOSM I'll set 1600 or 3200, but on my 5D3 I'll use 3200 or 6400.

Cheers

Garry