Messages

nrzi_decode(0x7F1) = 1374, FPS timer B is 1380 by default, so the values look alright. Also, if ML displays 1/30, 1/60 and 1/7800, it shows the right thing.

So... where's the problem?

Yes. Press the delete button to open ML menu and see what's green

This can be a hint in the intervalometer menu (where it already gives advice on how to avoid flicker, how many shots you will get and so on).

What are the register values at 1/fps and 1/8000? (or 1/4000, whatever you can get)

Yeah, good luck reprogramming the sensor timers to get that

Or, record at 5fp, group every 6 frames in post and add them. You got 30fps.

Heh, the regs are the same but trying it I just get shutter locked at 1/30. It does follow FPS tho... 28fps produces 1/28, etc.

The shutter blanking is used to compute the sensor duty cycle. Of course, if nothing changes there, it will consider duty cycle constant and will multiply it by 1/FPS (so it will follow the FPS).

Normal values (after NRZI decode) should be between 0 and FPS timer B - 1 (roughly).

You have checked shutter speed with expo override, right?

The register seems to be changing correctly in the logs though.

You may want to review this first: http://www.magiclantern.fm/forum/index.php?topic=4386.0

You mean, for animations?

I would keep the EDMACs for image processing. I actually prefer a GUI without flashing effects, even if it may look a bit old school. For current menu, plain memcpy with CPU is very fast, and far from being the bottleneck.

I've had no luck finding the address of the register in other cameras.. I can't even find it on 5D3.. I see the XX5F and XX61 which look like they have shutter but either get 1/0 when I use those or just one immobile 1/fps shutter. 5E and 60 addresses from 5D3 don't work.

For 5D2:
#define FRAME_SHUTTER_BLANKING_ZOOM   (*(uint16_t*)0x404B4110) // ADTG register 105F
#define FRAME_SHUTTER_BLANKING_NOZOOM (*(uint16_t*)0x404B4114) // ADTG register 1061

160, 200 and 250 are 100% identical in raw. And probably 400 HTP too.

As expected, this comparison simply tells the obvious: ISO 200 has cleaner shadows than ISO 100, but 1 less stop of highlight detail.

ISO 200 HTP is nonlinear digital gain applied to ISO 100. To compare them in raw, try ISO 100 without HTP and ISO 200 with HTP. These two use the same analog amplification.

In my tests I did not notice any difference between these two.

Can you post the raw files?

In theory, there's no difference between theory and practice. In practice, there is.

If you will ever read the FAQ, you will notice this: "Magic Lantern was created by reverse engineering an undocumented system..."

Yeah, nobody forced you to use ML.

Then you are using the wrong tool. Try the intervalometer and bulb timer.

Audio sync beep is fine (+/- a few frames). Wav recording is not. There are a lot of reports that it goes out of sync.

There are no cameras where it's working well (it goes out of sync in unpredictable ways). It's very hard to fix, we need to understand better the ASIF DMA.

I prefer to enable only things that are working well

There were no changes relevant to this.

Can you record a video showing the camera screen?

Nope, the limit is around 5 seconds on most cameras.

Overriding the shutter speed via ADTG may help. The sensor maybe just doesn't like a duty cycle of almost 100% (which is what you get with low-light FPS override).

You can compute exact shutter speeds from ADTG blanking register. No need to guess TG_FREQ_SHUTTER - you get the sensor duty cycle (0-100%), scale it with 1/FPS and that's it.

(and this proves that my previous hypothesis about NTSC TG_FREQ_SHUTTER was wrong on all cameras).

On current cameras, it's not possible to remap the scrollwheel without bringing some Canon menu first (in shooting mode it doesn't behave like normal buttons; instead of button press events, you get events about shutter/aperture/exposure changing). The only exception is the 5D classic.

I didn't look in the SL100, but it's unlikely that anything changed on this side.

The sound recording option from FPS submenu is for H.264...

Yes, it requires changing the file format.

The same trick can be used on 5D3 (recording on CF and SD at the same time, with a 20MB/s theoretical speed boost).