Messages

Yes, I smell that much is unused, but I don't know how to allocate it yet.

Forget 10 and 12 bits.

Back then, when we thought raw video was not even possible on 1000x cards at 24 fps, Luke Neumann asked me that even if I could record 3 seconds of video, he could definitely use that.

The 60D records 118 frames at 1734x976 24p according to my simulation. That's 5 seconds of video.

If we squeeze another 128MB of memory, that will be 184 frames (7.5 seconds). Not bad for short films IMO.

With end trigger and 400MB buffer, it will shoot 140 frames (~6 seconds). With current buffer configuration (280MB), you can do 96 frames (4 seconds) with end trigger.

Note: end trigger is not yet implemented, this is just theory.

Right, this will only benefit H.264.

For recording, the algorithm would be quite a bit different IMO:

- don't try large steps (photo ETTR does that to converge quickly)
- smooth transitions (can be done with shutter speed or digital ISO, or maybe just by enabling gradual exposure)
- override settings with the HDR video method (per-frame), not via properties (where timing can't be controlled)
- deflicker in post (with the same algorithm used for post deflicker, but applied in the converter)

It's totally possible, but not very straightforward.

Timing is very difficult, lookup those bugs (I guess there are videos showing them). But once you figure out how to sync with LiveView at scanline level, it's possible.

I don't know if the CPU is even fast enough for that kind of sync.

The increments are very fine (4096 = 0EV). It's applied before picture styles, just like digital ISO from the menu.

Did you use ISO 160 multiples? At these ISOs, the white level is lower than what's declared in dcraw source code, and I'm not sure ufraw takes care of this.

Can you upload a CR2 file?

Are you sure? did you try every single option from the Q menu?

Most probably you'll get the LED blinking continuously at startup and you'll wonder why

I'm sure you can, just as with all other Canon cameras

Just take out the SD card.

Yeah, but you can use Canon menu for that, don't see why it should be added in ML...

Note: you can't adjust analog ISO, only the digital one. It would help only a small amount with H.264 and no advantage at all with raw.

Initialization is done at startup, because config file is read before starting user tasks, and they perform initialization before entering the main loop.

Also, the updating logic is done in menu callbacks (whenever you change something from menu).

Either the DMA gets overloaded by larger frame sizes, or FIO_WriteFile returns a bit earlier (so ML overestimates the speed a little).

Playback frame skiping! Fast preview. Dynamic interface coordinates. Mask box updated to recording frame, now with no flicker. Restored frame skipping. Some module credits.

Source code please.

FYI: playback can be done realtime without skipping anything.

Why not just enable the intervalometer and silent pics? You get DNGs straight from the camera.

It looks like there are, I've counted them with grep. A few of them are stuff like battery discharge rate or measured card write speed, but most of them are menu entries that you can toggle.

Another obvious example: raw video. Enabling it needs active setup (putting LiveView into raw mode).

I did not count stuff that is saved directly in Canon memory (Kelvin WB, HTP, exposure values, picture style settings).

Who's going to review all those 100 or 200 500 settings and blacklist them?

g3gg0 already introduced separate config files for each module btw. Though, for performance reasons, having everything in one file might scale better.

Just add the kelvin value to flexinfo if it isn't already there.

If you have to go back to Auto, ML is probably not for you

No. As long as you record in movie mode (not photo!), the exposure values will be the ones displayed by ML (not by Canon!)

Can you modify an XML file so it contains that tag?

I don't have a license for Adobe software, sorry.

Well, that has to be deleted.

Actually I've tried it at some point, but was quite unstable.

The problem is that some settings also need some actions to be performed when you toggle them (think about audio stuff, you need to reconfigure the chip with the new settings, it's not enough to rewrite the variables).

This gets more complicated because we have more config file (magic.cfg, menu.cfg and a separate config file for each module). Maybe a folder for each mode, under "ML/SETTINGS", can do a better job.

Use movie mode and read the numbers from the screen.

If you turn off ExpSim, the camera will ignore the exposure parameters that you have set.