Menu

Show posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Show posts Menu

Messages - Marsu42

#1
Oh not, I've been using the old 7d1 for some time now to expand dynamic range of the old-ish sensor, and just discovered that with 800/100 setting there are still stripes after cr2hdr processing :-( ... obviously the algo has trouble telling noise from stripes. I've even discovered some shots from my good ol' 60d affected by this problem.

Strangely,  the old 14 bit algo does better, leaving more of the 7d1's vertical noise stripes in the sky instead of dual_iso's horizontal stripes. But even the old algo doesn't do as I'd expect it to.

... 14bit:


... 20bit:


I'll pm the source cr2 to alex or anyone wanting to help, pretty please, don't tell me these shots are broken :-(((
#2
Quote from: a1ex on October 16, 2016, 11:16:01 AM
Yes, pattern noise increases the stdev as well. Do you have some sample images to show the difference?

Well, I can send you two random 100/800 shots from the 60d and 7d1 that show the different "theoretical dr" line. But as my 60d is currently broken (I'll have someone look at the fuses, but that'll take time) I can only give you real "requested" test images from the 7d1.

What would be interesting to know which one of all the crop sensors before the 80d is how good for dual_iso. Sure, maybe the difference is minor and maybe 0,5ev, but if someone like me really wants Canon crop with enhanced dr and doesn't really care which old model to buy, it might matter.

Again, only looking at the sensorgen/dxo data the 7d1 should do best... even though it has a bad rep b/c of the pattern noise (which is on 5d2, too). I haven't tested how much the pattern noise affects dual_iso and the amaze algorithm, as it only appears visibly after pushing the shadows a *lot* which shouldn't be necessary using dual_iso.

Quote from: a1ex on October 16, 2016, 11:16:01 AM
The 20-bit version is merged.

Wupps, I missed that :-o

Quote from: a1ex on October 16, 2016, 11:16:01 AM
Are you asking about those few changes from here? These are not merged because they didn't pass my test suite.

I didn't try these either, but I'm keeping all my original non-cr2hdr'ed shots around in case some improvements are made :-)
#3
Quote from: a1ex on October 16, 2016, 10:11:12 AM
Protection status should be the same as DOS read-only attribute, so I guess it's easy to read.

Sounds promising, b/c the "quickly protect good images and delete the rest" method is the emergency rescue if you're running out of card space. But in  these cases, you're often pressed for time and don't really want to press trash/set for a couple of minutes :-\
#4
Quote from: a1ex on October 16, 2016, 10:18:41 AM
It looks at the optical black area (see dcraw -4 -E) and computes the noise levels from there, assuming that's the read noise. [...] Note the stdev is heavily influenced by hot pixels. Maybe I should replace it with some robust statistic.

So basically, this means the dr given by cr2hdr doesn't really reflect the actual dr of the image. No big deal as the picture itself is important, but it's a puzzler b/c the read noise spec from sensorgen of the 7d @ iso 100 is way lower than the 60d, but in cr2hdr it's the other way round. The 7d is said to have more "pattern" noise, maybe that is an influence on the stdev, too, like hot pixels?

Btw: What's up with the 20bit version, I'm using that one for ages, but it isn't merged. Was the reason that it's a lot slower but doesn't show reasonable iq gain from the 16bit version, or are there any other problems left with 20bit?
#5
Modules Development / Re: DotTune AFMA (dot_tune.mo)
October 16, 2016, 12:30:12 AM
Quote from: Manofmayo on August 11, 2016, 08:45:32 PM
I did the same thing with the Canon 35 IS and the Canon 50 1.4 and had a much large range of values (mostly between -10 and +10).   Does this make sense?

Nope - with my dot_tune experiments, everything worked as expected and the differences were in the +-2-range. The main source of error course is the distance to the target, but if that's fixed it should work.
#6
Quote from: ShootMeAlready on October 05, 2016, 05:48:07 PM
If all youre concerned about is still shooting, then I would agree with you that there are better DSLRs with ML release out there. The 7D has similar focus system as the 70D, and it has the key ML still shooter features you want.

Thanks for the assessment, I indeed now settled for a min-condition used 7d1 for less than half the price of a cashback 80d and strangely still way below for what a used 70d sells. True, a new camera has got warranty, but it's no good if you reach the shutter's spec'ed eol in no time with these frame rates :-\

The 7d1 has full ml, a sturdier shutter and build quality, and crucial af features that Canon, being Canon, has cut from the 70d and 80d: af point expansion and spot af. Esp. b/c the 70d sensor iq is basically what the 7d1 is, I really don't see the 70d as a still shooter's choice.

Quote from: treecreeper on October 11, 2016, 08:59:43 PM
I have been using the Beta for quite a few months and have not experienced any issues. One thing I am keen to try is the focus stacking option. What is the status of that feature? Is there a way for me to test or try it out?

If you'd venture a few posts below, you'd read that all the focus stuff isn't working atm. Or see https://builds.magiclantern.fm/features.html
#7
Quote from: Ant123 on October 14, 2016, 03:39:52 PM
Some progress with converting 40D port to 450D

Oh my, it isn't photoshopped? If not: good to see some people are not going along with planned obsolescence and try to make use of the good ol' gear!

Having said that, I personally wouldn't go below the long-running 18mp sensor for general photography nowadays, there are so many cams around with it and ML support so they should be more than affordable... I just bought a used 7d1. That's b/c I had a 450d-class cam (Nikon 3100) for some weeks on loan and wouldn't want to touch that usability and sensor iq again even with a 10-foot pole.
#8
Feature request: Delete all un/zero-rated and not protected files from directory.

Rationale: When shooting bursts, I often (pre)-select the pictures by rating them or at least protecting the ones I want to review further. Problem is: How to get rid of the rest? At home, I've written a script only moves all rated/protected files from the card. But with only the camera, I'd like to have a function to delete all at once except endless clicking trash->set or set+erase ... which is also prone to human with rated, but not protected files.

Any volunteers to add this to the file browser (Alex :-))? Would it even be possible, b/c ML of course would need to be able to check for the existence of a rating or the protection status (i.e. if the file is write protected or not)?
#9
Concerning the dual_iso dynamic range: Where does cr2hdr get the data to calculate the theoretical dynamic range from? I had a glance at the source code, and it seems to have to do with black/white-points and some mysterious "dark noise", but I don't understand it.

I'm asking b/c the 7d has higher dynamic range than the 60d according to dxo data, but the output of cr2hdr shows a little less. The resulting difference is nothing to write home about, even a 5d2 isn't a lot different at 100/800 according to cr2hdr output (but full frame can be used at 100/1600 while crop falls apart).

http://sensorgen.info/CanonEOS-7D.html
http://sensorgen.info/CanonEOS-60D.html

But I'm wondering: Why doesn't dxo's nominal higher dr @100 plus highrer dr @ 800 add up to a better dual_iso result than on 60d? Is it b/c of the higher noise of the older 7d1 (it has more "pattern" noise, too)? Or is the photographic reality for preventing clipped highlights and raising shadows completely different from what cr2hdr says anyway?
#10
Quote from: Armetz on October 04, 2016, 02:51:04 PM1. some merged frames differ quite much in brightness

I browsed the files in your merged "dual" folder and everything's fine and as it's supposed to be :-o

Quote from: Armetz on October 04, 2016, 02:51:04 PMthe merged files appear to have a color temperature shift towards blue

The wb of dual_iso is dodgy, you need to set a manual wb which is no problem as it's lossless on raw data.
#11
Camera-specific Development / Re: Canon 60D
October 15, 2016, 03:11:06 PM
My good ol' 60d recently made a motor sound at shill shooting it definitely shouldn't make and then died ... well, it has 330k shutter cycles. Now it doesn't do anything when turned on and I wonder if anyone knows how to diag what's wrong with it.

Does "doesn't turn display on" mean it's completely dead, like damaged a circuit board? Or could have a broken shutter or shutter motor have the same effect'?

I'm asking b/c there are rather cheap parts available for the 60d from China (either just the shutter blades or the shutter assembly), and I wonder if trying to fix it with the help of a friend of mine is worth it. With a new shutter, the one moving part would be replaced, everything else is fine.
#12
Camera-specific Development / Re: Canon 7D
October 15, 2016, 02:55:51 PM
Yay, I've just bought a reasonably priced near-mint condition 7d1 to replace my 60d which failed with 330k shutter cycles. Are there any active devs left on the 7d1 except for Alex, or has the world moved on by now?

General comment: I have to say I'm impressed, the 7d1 has everything I miss from the 80d for my photography (sturdy shutter, af point expansion, spot af) and most modern updates are jpeg- or movie-centered an don't matter to a still raw shooter like me. Only the lack of suport for rt flashes is unfortunate. As for the sensor, dual_iso fixes the low dr, for most of my lenses 18mp is 'nuff and if I'd want to have better iq I'd rather put my money into a 2nd full frame camera like a used 5d2/3. Or wait for 5d4's dual pixel tech with bokeh and focus point shift to make it into the lower realms.

Quote from: Macke1 on October 01, 2016, 08:27:02 PM
Ive just bought a 7d markii and i didn´t realise that ML wont work?? is it impossible or will it going to work some day?

This is the 7d1 thread, someone please change the title, here's the 7d2 one: http://www.magiclantern.fm/forum/index.php?topic=13746.0

Quote from: dawn_to_dusk_ on August 18, 2016, 01:52:28 AM
My question is 7D203 means 7D software version 203 or is a firmware to canon 7D mk2?

Even a minor fw update with just the Korean strings changed needs someone to find all hardcoded stub addresses in the Canon fw again which is a boring and lengthy pita. For what is changed this is definitely not worth it, just downgrade to 2.0.3

Firmware Version 2.0.5 incorporates the following fix.
1. Fixes a phenomenon in which the image files cannot be transferred using the FTP protocol via USB cable after the EOS 7D camera has established a wireless connection to the Wireless File Transmitter WFT-E5 (A/B/C/D).
#13
Camera-specific Development / Re: Canon 80D
October 11, 2016, 01:36:19 PM
Quote from: ccs86 on October 10, 2016, 04:08:08 AM
Take it to PMs guys.

And you're using the forum to tell us to pm instead of sending us a pm yourself, Mr. spare-time mod? :-)

Anyway, this tech discussion is an important preliminary for supporting ML on the 80d as it details features found in this model vs. previous or other current d6 models, thus outlining the fw changes to be expected while looking for stubs.
#14
Camera-specific Development / Re: Canon 80D
October 08, 2016, 03:26:00 PM
Quote from: Mr.Click on October 07, 2016, 08:07:55 PM
No you´re wrong. All AF points work as cross sensors even with a slow f5,6 lens , but only the mid Sensor works as a dual cross Sensor with a fast lens ( f 0..-f 2,8)
The Mid AF Sensor even works at f8 . With Special lenses & Converters even 27 AF Sensors work at f8.

I didn't ask about if it's cross or not, but how the *precision* of the f5.6 points with a lens faster than f5.6 is (like the mentioned 50mm prime). This is a real question, b/c the specs don't tell the whole story - the outer cross points a) could work quite well even with a fast lens, or b) could be totally worthless making the 80d a one-point-center-only camera with fast lenses. I have tested the 80d in a store, but only with the kit lens, that's why I'm asking other actual 80d users.

Btw the new motor-operated mirror makes me worry. With the older spring-operated mirror I got 330k shutter cycles out of my 60d, but I very much doubt a 80d will live that long if a motor is involved. Another case of "planned obsolescence"? Atm I'm rather thinking about getting a sturdy 7d1 with full magic lantern support right now :-o

My other general 80d comment of the day: Canon again put a lot of work into withholding features as the most useful af mode is missing ... "af pt expansion" is only on 7d2 and 5d3. So for precise af, it's one-point only with the 80d b/c the 9-point mode covers a much too large area :-\

QuotePlease excuse my worse English ;-)

Always a pleasure to exchange bad English with another German :-p
#15
Camera-specific Development / Re: Canon 80D
October 06, 2016, 05:08:49 PM
An off-dev-topic question to you 80d owners: The spec say that all af pts except the center one are only f5.6-precise.

I have some fast lenses, say a 50/1.8. Does this in *reality* mean that such a lens is completely unusable except with the center pt when wider open (like f2.5), or are the outer pts better than advertised, or is the af pt expansion so good it makes up for the lack of precision? That's b/c even with the 60d's center point (f2.8-precise) the af performance of this lens is dubious at best.

Thanks for actual experiences with this camera, I'm still not sure if I should get it or stick with an older and cheaper one that runs ml right now and can recover the lack of dr with dual_iso.
#16
Camera-specific Development / Re: Canon 80D
October 06, 2016, 04:58:18 PM
Quote from: a1ex link=topic=17360.msg172969#msg172969
date=1475668950
Yes, but that comes at the cost of 4EV of aliased highlights at 100/1600

Well, looking at my personal photography the highlights are usually in the sky so I couldn't care less about aliasing there, it's just important that it isn't blown. ymmv.

QuoteAnd I somehow doubt the shadows at ISO 100/1600 will actually look cleaner than at ISO 100, because of the half resolution

That's probably true, and my experience with current dual_iso, too. But again, for it it's about *any* recognizable data other than plain noise being there. For example viewers won't look at the deep black fur parts of an animal in detail, but it is disturbing if some black parts look either clipped or like pushed plain shadow noise. ymmv if you don't shoot moving black buffaloes in bright back lighting at noon :-)

Quote(which, in theory, reduces the DR by 0.5 stops and also affects the noise structure).

You can translate the loss of resolution into dynamic range? That sound kind of strange, or is this some mathematic wizardry about noise?

QuoteIf we have a bracketed image set, one at ISO 100 and another at ISO 1600, we can simulate a dual ISO image (lookup fake_dual_iso.exe) and see exactly if there is any extra shadow detail or not, compared to ISO 100.

Well, but you'd have to do this with actual 80d shots, right? If any 80d owner reads this, please do go ahead and shoot a grey patch @iso 100/1600 and run it through fake_dual_iso and let's compare with a pushed and bracketed-hdr shot.

QuoteInteresting that 5D4 contains ADTG strings on the AE processor, but none of them is present on the main processor.

Which means 5d4 users are screwed, but there's hope for 80d and 760d, right?

That would serve 'em right, b/c remembering the discussion about the 1d, a camera in the price range of the 5d4 isn't exactly covered by "poor enthusiasts adding features to their budget and midrange cameras" :-o ...

... if some other camera should be covered by ml, it'd be the 1d4 which should be about what the 7d1 tech is ... but the 1d4 is really cheap by now. But don't let Canon hear :-p
#17
Camera-specific Development / Re: Canon 80D
October 05, 2016, 12:53:39 PM
Quote from: eduperez on October 05, 2016, 07:40:33 AM
With that camera, in most situations you can just take the shot at ISO 100 and then raise the shadows during post-processing.

Riiiight, so I understand with dual_iso the camera would basically capture the same information at both iso levels, minus the dynamic range that is clipped at higher iso values of every other scanline?

Quote from: Greg on October 05, 2016, 01:36:18 AMLoss of detail for 1EV more does not make sense.

I'd tend to disagree there, 1ev is double or half the light which is a lot. And it can be exactly the dynamic range that is clipped from the bright sky or detail drowned in deep shadows (like the eye of an animal, even though the rest of the scene is brighter). That's the hassle with digital photography - missing data is missing data, but a little data can be still recovered and worked with. Plus the afaik dual_iso's "loss of detail" only affects the semi-over/underexposed areas, so by definition with the 80d's little ev gain it would be little to worry about?

Anyway, I guess either the current dual_iso code works with the 80d or it doesn't b/c Canon has done some radical changes ... though knowing Canon, it's more likely everything is the same as before but the marketing brochure got a face-lift :->
#18
Camera-specific Development / Re: Canon 80D
October 04, 2016, 11:59:24 PM
Quote from: a1ex on October 04, 2016, 01:00:42 PMEven if dual iso will work (which we don't know yet), the gain will be small, probably about 0.5 stops at 100/1600. From DxO data, I expect the 80D sensor at ISO 100 to be nearly as good as the 70D at say 100/800, without the resolution loss.

Ugh!? How so? I understand the dr gain of the 80d is achieved by eliminating low-iso read noise, so higher iso values are just as they used to be - but still, an overlap should get us more than 0.5ev? https://www.dxomark.com/Cameras/Canon/EOS-80D---Measurements

And if dual_iso on 80d isn't really something to look forward to, it means even a ml'ed Canon doesn't reach recent Sonikon at all: https://www.dxomark.com/Cameras/Compare/Side-by-side/Canon-EOS-80D-versus-Nikon-D7200-versus-Canon-EOS-70D___1076_1020_895
#19
Quote from: nikfreak on October 04, 2016, 02:09:48 PM
70D needs some unique trading and I admit that (if you got time) helping hands would be useful. Handling different revisions of a firmware number (1.1.1) didn't make it easy in terms of bug reporting and "valueing" user reports.

FPS override, dualiso in video mode and "raw zebras when quickreviewing after a shot" are broken atm. Being first cam with dualpixel AF didn't make it easy on 70D. As stated in 1st post most focus features don't work atm.

Thanks alex & nikfreak ... reading "cache hacks" and "low level re" doesn't make me feel like the solution is right around the corner, esp. if newer and more promising digic6 bodies will get more attention in the future. Afaik, these hacks were considered dodgy at best in the past, and if mpu reverse engineering would be realistic before hell freezes over alex would have accessed that $!#%& viewfinder and phase af system by now :-p

I don't currently have a 70d, and if it's missing 2/3 of the vital features I want to use I guess I'll either go straight to a 80d (that isn't much more expensive, go figure) or stick with a 100% supported camera like 60d or 6d :-o. Using these, it's easy to forget that ML isn't supported by Canon and it's a sheer miracle all these features used to be working...



#20
Camera-specific Development / Re: Canon 80D
October 04, 2016, 12:06:47 PM
Quote from: eduperez on September 01, 2016, 02:53:18 PM
If you are thinking about buying a new camera and you positively need ML, do not buy one not supported now with hopes that it will be supported in the future; it may never be supported, and you will be stuck to an unsupported camera.

... or something in between. As ML mostly relies on fw hooks or accessible variables, and there's no telling what Canon cooked up again, it's not certain what ML features will work or won't work ever. Combining the expanded dynamic range of the 80d with dual_iso would be a dream and beat Sonikon, but notice the "but".

So if you need something specific even a general "Look at qemu, ML will run on camera xyz sooner or later" announcement isn't enough to go ahead and purchase it.
#21
Quote from: nikfreak on October 04, 2016, 12:16:06 AM
What's not working
* Only fast zebras work (no raw zebras)
* Trap focus only works in photo mode (not Liveview). All other focus features are disabled (follow, rack, ...)
* Dualiso only works in photo mode (Looks like we are stuck in movie mode atm like 50D / 7D)

@all current 70d users and devs active on 70d:

I'm thinking about getting a used 70d as a replacement for my 60d that just broke (well, 330k shutter cycles...) as I don't see ML on 80d anytime soon. I don't fully understand ML's status on the 70d and the prospect of the features, this not getting merged doesn't feel good :-\ ...

1. raw histogram - I take it at least this works on 70d even though raw zebras don't?

2. rack focus for focus stacking - is this completely broken on 70d or is there still hope (lens.c: "focus stacking is still buggy and takes 1 behind and 1 before all others afterwards are before at the same position no matter what's set in menu")?

3. dual_iso - is this working 100% in photo mode, or are there any unexpected problems to be expected?

I did some ML development on 60d and 6d, but with the latter I was frustrated by things that simply won't work "ever" like focus patterns - that's why I'm asking if there's active development on 70d other than adapting the stubs for a new fw version.
#22
Quote from: Danne on October 04, 2016, 06:01:40 AM
These are the tags which seems necessary for correct white balance(check below). The other tags mentioned before aren,t needed. I,m trying to fetch CameraCalibration tags in CR2 but somehow it only shows when converted in adobe dng converter. If anyone have any clues here?

Doh - that's closed source for you. There have been attempts to get the correct wb from tags here before, all doomed b/c obviously we lack the secret sauce ingredient :-\

Adobe seems to have managed to make use of the maker notes, probably using some insight straight from Canon we don't have. It might even be that there's no single algorithm to convert the values, but actually a model-specific lookup table as for new cameras there has to be a new acr version.
#23
Quote from: josepvm on October 02, 2016, 02:38:46 PM
Chromatic aberration is a defect of the lens. How could Dual-ISO introduce chromatic aberration?

In theory, I guess it's possible that cr2hdr confuses that ca-reduction algorithm of some postprocessing software which are tuned for specific, untouched raw files, i.e. dual_iso shots could have more ca left. However, I have never observed this effect myself (using acr) even with cheap fast lenses, but I didn't try to provoke it either to prove or disprove the possibility.

Quote from: josepvm on October 02, 2016, 02:38:46 PMSee also http://www.magiclantern.fm/forum/index.php?topic=10265 (in particular, posts from ayshih and Marsu42)

Good advice, that :-p ... I faintly remember me stating that the color reproduction of very different iso levels (like iso100/iso1600 on a crop camera) is a source of confusion in any case, i.e. you cannot expect this to behave like a perfect iso100 bracketing-hdr shot with expanded dynamic range.

Alas, I admit I was nitpicking when I started using dual_iso, but since I do nature shots and no one except me knows how the original scene looked anyway a perfect color reproduction or wb doesn't matter that much to me. When in doubt about the wb, the "auto" setting in acr (i.e. Lightroom or Photoshop) works just fine as a starting point, and not getting blown whites or drowned blacks is more valuable than anything else.
#24
Quote from: xorpi on September 30, 2016, 06:52:58 PM
Dual ISO with actual performance is not good for any types of hobby or professional photography...

This is just FUD. I'm using dual_iso since it was released and use it on most outdoor shots nowadays. When in doubt, a little wasted dr and semi-clipping with dual_iso beats completely blown highlights and low-res blacks drowned in noise.

With recent Canon sensors ml supports (like 6d) you get excellent iq, 16bit raw files and can actually match the dynamic range of the scene and the image by selecting the correct iso spread (100/200, 100/400, ...) = optimal bit-res usage. Even on older sensors like the 18mp crop the shadow noise is ok up to 100/800, but with hugely improved dr.

As you need to back up your shots anyway, there's no additional file space requirement (back up the dual_iso dng, work on the cr2hdr-processed dng). The only hassle is the broken auto-wb, the striped preview and the time to process the files.
#25
Quote from: bart1983 on September 24, 2015, 01:24:03 PM
Maybe it would work with short HalfShutter impulses every 10 Minutes?

In that case, did you try to merge the recent lua scripting branch and use a script that does this? I know it involves some work atm, but this sounds like a prime example for simple user scripting "press button x, wait y, repeat".