Full-resolution silent pictures (silent.mo)

Started by a1ex, July 01, 2014, 05:11:15 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

JoshuOne

Greetings, I am attempting FRSP with intervalometer on my 5D2 and I am only getting one picture with an ugly black bar across the pic with "intervalometer wating for first pic" across it. Please help, I am using the latest nightly, I have tried with IV set to take photo on half shutter and on first pic, I have the IV set to like ten pics, the shutter is at like 1/15 of a sec, I have tried with canon menu review time like 2 sec and off, expo override on and off, MLV and DNG... I imagine there is some small thing I am missing, not a noob to ML, but to FRSP I guess I am. Could someone who is successfully setting up FRSP with intervalometer please let me know what I might be missing? Thanks to all the devs!!! a1ex g33o mk1174 and everyone!!

Licaon_Kter

Replied to the commit but I'll add it here too:

On my M1 I now get:

* image on screen for X sec followed by black screen that waits for a button press with Canon->"Image review"->"X sec" and no matter what ML->"Image Review Settings"
* image on screen that waits for a button press with Canon->"Image review"->"Hold" and no matter what ML->"Image Review Settings"
* black screen that waits for a button press with Canon->"Image review"->OFF and no matter what ML->"Image Review Settings"

What settings should I set and how to get NO image review at all BUT no black screen waiting for a button press either?


a1ex

Correct - this behavior minimizes the power draw for an external-triggered timelapse.

To get no image review at all, and return to LiveView right away... I don't know, press half-shutter twice?

itsskin

Quote from: Audionut on April 16, 2015, 07:25:30 AM
Underexposed.  Increase exposure and/or ISO.

Sure, and blow all highlights :D Blame ETTR for exposure :D

Licaon_Kter

Yeah ticky.

Trying to record a bit with the intervalometer, to answer @JoshuOne, got the timer running, 10s, FRSP, MLV, the screen was always off which is nice, but after 9 frames or so:
ML ASSERT:
lv
at ../../src/zebra.c:4202 (idle_display_dim), task cls_task
lv:0 mode:3


Magic Lantern version : Nightly.2015Apr15.EOSM202
Mercurial changeset   : 693ec08e528e (unified) tip
Built on 2015-04-14 22:08:34 UTC by [email protected].
Free Memory  : 134K + 3233K

and the screen looked like this (had to pull battery for it to recover):


Also, I initially thought that it might be a problem with my builds, as the official nightlies did not exhibit it until now, but I get asserts like this all the time:
ML ASSERT:
lv
at ../../src/raw.c:1690 (raw_lv_request), task ettr_task
lv:0 mode:3


Magic Lantern version : Nightly.2015Apr15.EOSM202
Mercurial changeset   : 693ec08e528e (unified) tip
Built on 2015-04-14 22:08:34 UTC by [email protected].
Free Memory  : 140K + 3095K

and
ML ASSERT:
lv
at ../../src/raw.c:1690 (raw_lv_request), task livev_hiprio_task
lv:0 mode:3


Magic Lantern version : Nightly.2015Apr15.EOSM202
Mercurial changeset   : 693ec08e528e (unified) tip
Built on 2015-04-14 22:08:34 UTC by [email protected].
Free Memory  : 158K + 3135K


What do they mean?


Quote from: itsskin on April 16, 2015, 10:02:33 PM
Sure, and blow all highlights :D Blame ETTR for exposure :D
Yeah ETTR and FRSP don't yet cooperate.

Audionut

Quote from: itsskin on April 16, 2015, 10:02:33 PM
Sure, and blow all highlights :D Blame ETTR for exposure :D

No, always blame the user  :P

What highlights were you trying to protect?  Some stars?

dmilligan

Quote from: Licaon_Kter on April 16, 2015, 10:17:53 PM
What do they mean?

It means that ETTR tried to request LV raw data while LV wasn't actually running.

Licaon_Kter

I was testing FRSP so sometimes it ended up in a black screen hence ETTR/zebras could not get its needed data, makes sense.
For now I will set Zebras->RAW Zebras to OFF so that maybe that won't trigger again on intervalometer use, when the camera does not get a chance to display the image at all anyway.

Another thing, are asserts a sign of things being unstable (like the error that I got later) or just plain warnings that can be ignored?

dmilligan

It just depends on the actual assert itself. These types of asserts (asserts from ML code), indicate that something ML considers invalid happened. Asserts in general are a sort of a 'self-test' of the code. If something "shouldn't" happen, then we explicitly check and make sure that it doesn't, and if it does, throw an error, so we'll know about it and fix it (this is good programming practice). An assert could be for something simple with no real possibility of causing instability, or it could be for something quite dangerous, the assert itself could also be the problem (an invalid or incorrectly coded assert).

There are also asserts that come from canon code, these are a different sort of beast, since we didn't make them ourselves, don't have the original code and we can't easily tell what the assert is for and why (at least not without a lot of reverse engineering and analyzing assembly code).

PitifulInsect

Hi!

Way back here,
http://www.magiclantern.fm/forum/index.php?topic=12523.msg121962#msg121962
I understood that at capture time, FPS timer A controlled the time allotted per row.  Per that interpretation, some later posts talked about reducing the timer value for faster exposures.

Then here,
http://www.magiclantern.fm/forum/index.php?topic=12523.msg143510#msg143510
I got thrown about the comment that the sensor wasn't being read in rolling shutter mode.

I'm really interested to know if timer A can be increased for slit-scan full-resolution images.  Because that would be awesome.

I'm wondering if:

  • A - It's totally possible already, and someone will just put in the menu option for adjusting timer A if I ask nicely, or
  • B - This finding opens the door to full-resolution slit-scan but it will require some engineering, in which case I'll roll up my sleeves and contribute if/when I have something, or
  • C - I've got confused or lost at some point and there's no direct approach to this idea yet

Thx :-)

a1ex

I didn't really experiment much with this timer, but I couldn't decrease it too much. You should however be able to increase it quite a bit, which will give a stronger gradient.

I don't know yet if the timer can be changed during an exposure - if so, you might be able to tweak the shape of the gradient (e.g. try to make it linear in EV space, rather than exposure time).

Once one figures out how to drive the sensor in fullres rolling shutter mode, slowing it down for slit-scan capture should be pretty easy.

On 500D, Greg was able to capture a full-width LiveView image (in rolling shutter mode) here with the help of FPS override.

JoshuOne

I must confess, I do not understand. Is tehre something special I must do in order to use intervalometer with Full Res Silent Pics on the 5Dmkii? Any help is much appreciated. I am in SF this weekend for a journalism conference and would love to be able to get some crowded Market street timelapses without burning out my tired shutter  :) As it is now, I get one photo with a black bar across it that says "intervalometer waiting for first pic..."
Thanks everyone.

dmilligan

Use the 'leave menu' trigger.

It appears that the 'take a pic' trigger only triggers off of an actual photo being taken not a silent picture (I suppose that should be fixed). The 'half shutter' trigger doesn't work so well either if you try and start from LV, as it conflicts with the silent picture itself (which also tries to trigger off of half shutter). When you press half shutter, it's intercepted by silent picture, and you simply get a silent picture and the intervalometer isn't started (it didn't get that you pressed half shutter and displays that message that you see, b/c it thinks you still haven't pressed half shutter). Half shutter trigger seemed to work fine though, if you started outside LV.

So, your best bet is to simply use the 'leave menu' option. It works just fine in any scenario.

JoshuOne

Thanks so much, dmilligan! That was one thing that I hadn't tried and the one thing that worked. I am not sure how well, but a simple 10-shot test seemed to do fine. THANKS!! :)

Ottoga

I can confirm that the "leave menu" works fine. Also, Even though your camera will likely be on a tripod, I would also set a start delay of a few seconds. Just a bit of insurance  to ensure that there is no camera shake due to a heavy handed button presses when exiting the menus.
EOS 7D.203, EFS 55-250mm, EF 75-300 III, Tamron 16-300 DiII VC PZD Macro, SpeedLite 580EX II.

Licaon_Kter

Quote from: Licaon_Kter on April 16, 2015, 10:17:53 PM
Trying to record a bit with the intervalometer, to answer @JoshuOne, got the timer running, 10s, FRSP, MLV, the screen was always off which is nice, but after 9 frames or so:
ML ASSERT:
lv
at ../../src/zebra.c:4202 (idle_display_dim), task cls_task
lv:0 mode:3


Figured it ( by setting up ML multiple times ) how it was trigger ( BUT it's obvious now since it's written RIGHT THERE in the assert daamit :P ) and reported it: https://bitbucket.org/hudson/magic-lantern/issue/2262/dim-display-triggers-assert-in-zebras-when

Levas

Does anyone know, perhaps A1ex  ???, if the sensor can be read out line by line, but with a pause somewhere.
Is it possible to create a real gradient filter by reading out the highest lines quick, then pause, and resume with the rest of the image lines ?


Slasheal

I'm using a 7D.
I tried old versions and latest released FRSP module.
I tried fresh magic lantern install.
I have activated expo sim & expo override.
In M mode.

I have 3 lens:

50mm 1.8
55 250 STM
10 18 STM


With the 50mm, setting the manual aperture in live view works great.
I can see direct result on screen.
The FRSP is exposed as is the simulation.

With my two STM lens, aperture change leads to.. Absolutely no change in exposure. I can see exposure result for 1/5sec when I halfpress the shutter and when "exposim" from canon menu is darkened.
But when the picture is taken with FRSP : exposure correspond to what I have on screen, so it's overexposed.
So I can't use FRSP in daylight because it's always over exposed.

Is this normal? Why is it behaving differently with my 50 and my STM lens?

I can make a video if you wish so.

DeafEyeJedi

It is normal because for FRSP the fastest shutter is avg @ 1/15th of a sec for the most part.

So either close down the Iris or get a ND for daytime use.
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

Slasheal

I know that I can't get a fast shutter speed.
I wasn't talking about the shutter speed.
You advise me to close the iris.
The iris as you say is also called the aperture.

And I just tried to explained that I can't change the aperture with my two STM lens while it works perfectly fine with my canon 50mm 1.8 with the exact same settings, just by changing the lens.

50mm: touch the aperture settings: exposure change on screen: exposure of the silent pic change.
My two new stm lens: touch the aperture settings: change on camera: no change on screen: no change on exposure.

Strange.
I tried with MF/AF and with IS/no IS just to see...
Can't get why it is stucked only with my two stm lens.

a1ex

Sounds like expo override may not work with STM lenses. Any other STM lens users can confirm?

Can you manage to take a picture with the DOF preview button pressed?

josepvm

I have a 18-55mm IS STM, a 55-250mm IS STM and a 24mm STM. Expo override works Ok on the three lenses, no problem to adjust aperture during FRSP on a 500D.

pholler

I can confirm that expo-override does not work on 5D3 + EF 40/2.8 STM in the latest nightly. I had the same issue yesterday. The aperture does not close more than about f/4.0 or f/5.6.
It works fine with other lenses.

pholler

I tried using FRSP with A-ETTR in a day-to-night timelapse two days ago with the actual nightly. For that purpose i used the .mlv-format, set A-ETTR to a max shutter-time of 10s and used a 12s interval. The result was that A-ETTR worked fine but i only got a picture every 24s and in the metadata the shutter time was always 1s but i can see from the exposure of the picture that the shutter time changes very often. So A-ETTR worked correct but the metadata was written wrong. Aperture and ISO worked fine.
Do you have any ideas about the shutter time? Is there something wrong in the code or a user-error?

Licaon_Kter

Quote from: a1ex on April 20, 2015, 11:04:52 PM
Sounds like expo override may not work with STM lenses. Any other STM lens users can confirm?
Confirmed on my EF-M 22mm 1:2 STM too.
Now that I think of it the other thing shows it exactly as described.

Quote from: a1ex on April 20, 2015, 11:04:52 PM
Can you manage to take a picture with the DOF preview button pressed?
Umm, having C.FnIV (6) setting Trash for "DoF preview", a single-press will trigger a quick iris closing up to my set up aperture AND after the iris moves back to max-aperture it will trigger a FRSP too, weird !? Also while *-press or half-shutter press exit the preview, pressing Trash again does not.
Hold-press Trash most of the time will close the iris properly but the screen will go from "Hold on..." to black screen with "Raw error" until you depress Trash and after it will stay as a black screen that waits for *-press or half-press, no FRSP will be taken on hold.

Actually as far as I can see it will always make a FRSP at max aperture no matter what I have selected and whether ExpOverride is used or not.