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.

edt

Just to add a bit more info.. The added audio functionality allowed for listening to the audio (using an extra lead in the av port ) and gave an on-screen live audio level when recording

Walter Schulz

60D features Digic 4 processor. 100D comes with Digic 5. See feature matrix -> Audio and get aware for the big gap between those cams coming with/without Digic 4. There is/was a new audio interface in development but - as always - there is no schedule. Digic 5 cam owners are stuck with audio functions provided by Canon.

IDA_ML

@edt

I am sorry, I misunderstood your question.

IDA_ML

I have a few questions to the developers working on the 100D;

1) Is it possible to fix the "hick-up" issue on the 100D in the 5x magnification modes which makes the camera unusable in these modes?  These are these very annoying jumps that occur at random places in the video clip where it jumps one frame back and then continues from there.  This phenomenon is not observed with the 7D for example.  If this "hick-up" issue could be fixed, this would allow the 100D to record short video clips with a very high quality at resolutions up to 2520x1080, (tested already).  Very often, especially when filming travel videos, this is necessary for landscapes, architecture or scenes with a lot of fine detail.  They simply look gorgeous when shot with the 100D at those resolutions.

2) Now that we have sound with the 8...11-bit and 12-bit lossless modes working continuously and synchronously at 1728x972 resolution and 24 fps, is it possible to add sound also to the 5x magnification modes where the above resolutions are possible?  This, of course, makes sense only if the "hick-up" problem is solved.

3) Right now, in-camera playback of 10 and 12-bit uncompressed MLV files is not possible.  This yields a black screen with the frame counter working.  I believe, this has something to do with the raw-tweak module not working properly.  Is a fix of that module for the 100D possible?
===================

For me personally, the small, 400 grams heavy 100D is a "dream come true" for travel videography, especially after the latest developments.  It is now very stable and all the MLV tools make it very convenient to use even at low light.  If the above three issues could be fixed, this will make this camera an extremely powerful high-mobility tool for RAW video recordings with sound - perfect for those who travel a lot. 

serek

When shoting in dual ISO I have a lot of pink colour in overexposed areas ::)

scherbakoff.dima

There is no actual experiments build (lua-fix)  for 100D :(
Is it something wrong with build bot configuration? Or is current branch broken for 100D build?
If someone has fresh build of ML, can you share it, please?
Thanks in advance:)

OlRivrRat

   Attempt to add 28Oct17 crop_rec.mo to Latest 16Dec17 Nightly causes Error.
ORR~DeanB  ~~  80D-ML  &  SL1+ML  &  5D2+ML  &  5DC+ML  &  70D+ML(AliveAgain)

dfort

@OlRivrRat - Of course that won't work. Why aren't you using the crop_rec_4k experimental build?

OlRivrRat

@ DFort ~ I have been until the 16dec17 Nightly happened. So I installed it & then decided to see if the Missing .mos could be added.
I'm just an appreciative user not 1 of the Wizards so I would have no way of knowing it would be an "of course that won't work"
ORR~DeanB  ~~  80D-ML  &  SL1+ML  &  5D2+ML  &  5DC+ML  &  70D+ML(AliveAgain)

dfort

Surprised it was working before. There has been changes in the back ends on most of the experimental builds and I believe that the crop_rec module is using some of those changes.

The crop_rec module for the 100D gives you a wide aspect ratio image at 50/60fps using a 3x3 pixel pattern which reduces aliasing when compared to mv720 mode.

In any case, why not use the crop_rec_4k build for the 100D? That build adds 10bit/12bit and lossless compression which will allow you to use a much larger image size and/or frame rate than regular 14bit raw.

IDA_ML

Hello everybody and a Happy New Year to all of you!

Over the holidays, I finally found some time to test the latest experimental builds for the 100D.  Here is what I found.

100D Dec. 19-th Experimental build:
===================================

1) The RAW-tweak module does not work in any of the RAW-video modes.  When the record button is pressed, the "Threads failed to start" message appears and the screen turns black.  A red dot and a "Starting" message appears next to the RAW-video mode selected and recording never starts.  When the module is disabled, recording works normally but when the RAW-tweak module is enabled again to playback the recorded clips, a black screen occurs.  When now halfshutter is pressed the red LED starts blinking but the screen remains black.  The red LED does not stop blinking even after the camera is turned off.  The only way out of this state is a battery pull.  So, I had to disable the RAW- tweak module to continue testing.

2) With the RAW-tweak module disabled, the RAW-video (MLV) mode works fine with sound at 1728x972 resolution in the normal (noncropped) mode.  The following recording times were measured at the 23,991 fps (23,976 setting).

   10-bit - 369 frames
   12-bit - 174 frames
   14-bit - 111 frames

The same behaviour is observed with the Movie crop mode which works with sound too.  In-camera playback is possible only with the 14-bit clips.  When playing back 10 and 12 bit clips, the screen turns black but the frame counter works which is useful for checking how many frames have been recorded.

The RAW-video (MLV) mode works with sound also in the 5x-magnification mode at resolutions up to 2496x1080 (high video quality).  Unfortunately, the hick-up issue (see below) makes this mode unusable.

3) With the RAW-tweak module disabled, the RAW-video mode (MLV-lite) works fine (no sound) at all bit rate modes in both:  the normal (uncropped) and the Movie crop modes.  Continuous recording at 1736x976 resolution is possible in the 8...11 and 12-bit losslessly (LL) compressed modes.  Even at 14-bit LL recording is quite long (between 240 and 400 frames with a normally lit scene dependent on the fine detail in the scene).

Movie crop mode works also in all bit rates at 1800x1008 resolution at 23,976 fps.  At 8...11bit LL even continuous recording at this resolution is possible.  At 12-bit LL, which is considered the sweet spot for LL video, recording lasts quite long (about 50 s with a well lit scene).  At 14-bits uncompressed where only about 2 s recordings are possible, just the first 20 frames are fine, the remaining ones are broken.

Also in the RAW-video mode (MLV-lite) the instability from #1 is observed, especially when switching from one bit rate to another.  When in-camera playback is attempted which works in the three LL modes and at 14-bit uncompressed, the screen very often but not always turns black, (hard to reproduce). When now halfshutter is pressed the red LED starts blinking but the screen remains black.  The red LED does not stop blinking even after the camera is turned off.  The only way out of this state is a battery pull.  This behavior occurs with the RAW-tweak module disabled, so it does not have anything to do with the RAW-tweak module.

4)  The most serious problem with this build as well as with older builds, (already reported several times in this and other threads), remains the "hick-up" issue at 5x-magnification mode where the 100D is capable of recording at resolutions up to 2520x1080 and quite reasonable recording times (I got 6 to 8 s recording in the 8...11-bit LL mode at 24 fps at that maximum resolution).  Hick-ups happen in both:   the  RAW-video (MLV) and the RAW-video modes and only at 5x-magnification, regardless of the frame rate and resolution set.  The hick-ups are jumps (one frame back and then two frames forward) that occur at random locations in the video and happen during recording.  They are also seen during in-camera playback but not in the preview mode before the record button is pressed.  During recording, they are not seen either.  Interestingly, the jumps are seen only on moving objects in the scene. The static ones don't jump.  I have made a test video with moving vehicles that illustrate the issue (see below).  Please note that the hick-ups are noticeable only with the tram and cars in motion while the static objects in the background (buildings, street, posts, etc.) don't jump.  This does not mean that the jumps are not there.  On static objects they are just not seen since there is no difference from frame to frame.  I have confirmed this by shooting clips on entirely static scenes with no moving objects and no hick-ups are noticeable.  Here is the download link of a hick-up video which will be active for 7 more days:

https://we.tl/9iEYdM8rky

My explanation to the hick-up problem in the 5x-magnification mode on the 100D is that this issue is related to frame synchronization (possibly vertical synchronization).  The 100D loses synchronization at random moments during recording in that mode.  The reasons why I think so are as follows:

1)  The hick-ups are not seen with static scenes where there is no difference from frame to frame;
2)  In scenes containing static and moving objects, the jumps are clearly seen only on the moving objects (see tram and cars in the above example) while the static ones stay free of jumps. 
3)  Sometimes, (hard to reproduce), when in-camera playback of a hick-up video is attempted, an earthquake motion right after the jump is observed which is very similar to the earthquake motion of older 7D builds before the RAW_Slurp issue was solved. 



A request to Dfort, Nikfreak, Danne and everyone who can read and write code and is willing to help:
=================================================================================

Could you guys please take a look and see if you can fix the hick-up issue in the 100D?  It is a pity that this highly capable little camera is crippled in one of its most creative modes - the 5x-magnification mode where RAW video recordings with amazing detail and very high quality are possible at resolutions up to 2520x1080 and no VAF filter is needed.  I have all reasons to believe that the issue has something to do with the RAW_Slurp and/or vertical synchronization.  Dfort, you were so successful with fixing the issue with the 7D.  Maybe, you could apply your experience to the 100D too. 

Thank you!

=================================================================================

IDA_ML

I also have a question to А1еx regarding the silent module on the 100D that I have been using a lot for timelapse work.  I have been investigating the possibility for using it with stabilized lenses for shooting single full-resolution silent pictures in the Life View mode without actuating the mechanical shutter.  This is one of my favorite ML features since it allows taking single DNG photos at the highest sensor resolution, also in the Dual ISO mode, at locations where shutter noise is highly undesirable (churches, museums, lectures, etc.).  An additional bonus is the fact that pictures are taken in the mirrorless mode, so vibrations due to mirror flipping are avoided.  Thus sharp handheld pictures are possible at shutter speeds as low as 1/25-th down to 1/4-th of a second with stabilized lenses. 

There is one problem, however.  I usually use the back button for precise focusing at 5x or even 10x-magnification.  In this process, the image stabilizer (IS) of the lens is active and a very precise focus can be adjusted even at low light. When I am ready with focusing and metering and want to take the picture by pressing halfshutter, the IS continues working only for about 1 s. after I release the back button and then it stops working.  It does not get activated when I press halfshutter to take the picture.  This means that at the moment I take the photo, I have no image stabilizer engaged and the picture turns out blurry due to camera shake.  Sometimes I manage to press halfshutter very quickly, within this 1 s. after I release the back button hoping that the IS is still working during the actual photo taking but the time is too short to avoid camera shake.  Moreover, in this short time it is impossible to go back to full-screen LifeView for proper framing and or leveling.

So, my question is:  Is it somehow possible to increase the IS time from 1s. to 3 or even 5 s. after the back button is released?  An even better option would be to reengage the IS upon halfshutter press and holding down and the picture is taken only in the moment, half shutter is released.  In this way, the photographer would have enough time for proper framing and taking a deep breath during IS operation, before the actual photo is taken.

Thank you.

dfort

Quote from: IDA_ML on January 02, 2018, 03:58:21 PM
Dfort, you were so successful with fixing the issue with the 7D.  Maybe, you could apply your experience to the 100D too. 

I wasn't successful at fixing the 7D hiccup issue. Took a break from that to deal with some focus pixels that are showing up on other cameras including the 100D.

12georgiadis

Ok, so I decided to buy one 100D to play with it. Is there a possibility to implement h264 proxy function (0.1x) to test it at low bitrate and see how is the recording behavior with 8...11bits in the same time ? This could allow realtime playback like with the 5DmkIII. This is really important when working with actors.

IDA_ML

Quote from: 12georgiadis on January 03, 2018, 02:40:19 PM
Is there a possibility to implement h264 proxy function (0.1x) to test it at low bitrate and see how is the recording behavior with 8...11bits in the same time?

I don't think so.  I never tested this function myself since I don't need it but I remember someone saying/writing that simultaneous h.264 proxy recording does not work on the 100D.   This would be too much for this little camera with only about 40 MB/s write speed. 

12georgiadis

Quote from: IDA_ML on January 03, 2018, 03:54:35 PM
I don't think so.  I never tested this function myself since I don't need it but I remember someone saying/writing that simultaneous h.264 proxy recording does not work on the 100D.   This would be too much for this little camera with only about 40 MB/s write speed.
Oh yeah, now I remember danne's post about an issue with h264 proxy (same as eos-m). A liveview problem, not bitrate issue. I think it could handle continuous rec within the 41MB/s limits with certain conditions (with 8bits and  underexposure) and with 0.1x bitrate. But we need to get rid of these glitches first of course.
@danne @IDA_ML, do you have the name of the build with h264 proxys or it is already there in last crop_rec_4k build ? Can we change the bitrate values ?


Envoyé de mon iPhone en utilisant Tapatalk

DeafEyeJedi

Quote from: IDA_ML on January 02, 2018, 05:00:45 PM
So, my question is:  Is it somehow possible to increase the IS time from 1s. to 3 or even 5 s. after the back button is released?  An even better option would be to reengage the IS upon halfshutter press and holding down and the picture is taken only in the moment, half shutter is released.  In this way, the photographer would have enough time for proper framing and taking a deep breath during IS operation, before the actual photo is taken.

I second this as well. Excellent question @IDA_ML!
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

IDA_ML

Quote from: 12georgiadis on January 03, 2018, 05:01:08 PM
do you have the name of the build with h264 proxys or it is already there in last crop_rec_4k build ? Can we change the bitrate values ?

I am using the Oct. 22 2017 build for the 100D and it has the H.264 proxy recording feature in the submenu of the RAW-recording function (8...11 bit LL).  When I turn proxy recording on and start recording, the camera crashes and needs a battery pull.  I told you, it doesn't work!

12georgiadis

Quote from: IDA_ML on January 04, 2018, 12:11:28 AM
I am using the Oct. 22 2017 build for the 100D and it has the H.264 proxy recording feature in the submenu of the RAW-recording function (8...11 bit LL).  When I turn proxy recording on and start recording, the camera crashes and needs a battery pull.  I told you, it doesn't work!

Thank you for the report IDA_ML. If the camera crashes, this could not be a bitrate issue but something else. It would be great if this very useful feature can Be fixed on this model to test the double record (0.1x h264+ 8...11bits) on stable conditions. This way, we'll see if 41MB/s is enough or not. I can't test with my 5Dmk3 because the SD port is 20MB/s so we can't compare (on CF card, with 14bits+h264 proxy, it stops but doesn't crash)


Envoyé de mon iPhone en utilisant Tapatalk

IDA_ML

Quote from: 12georgiadis on January 04, 2018, 12:56:19 AM
I can't test with my 5Dmk3 because the SD port is 20MB/s so we can't compare (on CF card, with 14bits+h264 proxy, it stops but doesn't crash)

Can you say at what camera settings exactly, the 5D3 does not stop or crash but records both - RAW video and proxy properly?  Maybe, this would be a good starting point for other cameras too.

12georgiadis

Quote from: IDA_ML on January 04, 2018, 10:34:24 AM
Can you say at what camera settings exactly, the 5D3 does not stop or crash but records both - RAW video and proxy properly?  Maybe, this would be a good starting point for other cameras too.

Tests with 5Dmk3, recording both raw + h264 proxy :

With Raw that recs on a Compact flash 1000x 64GB and h264 proxy on SD card :

Lossless 14 bits : continuous
Lossless 12 bits : continuous
Lossless 10 bits : continuous

Idem but raw + h264 proxy recording both on the compact flash :

Lossless 14 bits : continuous, sometimes stop in certain scenes
Lossless 12 bits : continuous
Lossless 10 bits : continuous


Envoyé de mon iPhone en utilisant Tapatalk

IDA_ML

And this is at what resolution?  Is the behavior the same at all resolutions and frame rates?

12georgiadis

Quote from: IDA_ML on January 04, 2018, 12:45:58 PM
And this is at what resolution?  Is the behavior the same at all resolutions and frame rates?
Resolution is 1920*1080 at 23,976fps.
I didn't try at 25fps but it may be the same results.


Envoyé de mon iPhone en utilisant Tapatalk

IDA_ML

Thank you very much, 12georgiadis!

nikfreak

@IDA_ML: I assume you enable fps override in 5x to extend the recording duration? if true please report back if the hickups also appear with fps override disabled
[size=8pt]70D.112 & 100D.101[/size]