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.


Messages - dyfid

Pages: [1] 2 3
1
General Chat / Re: GPL issues with ML post processing software
« on: September 21, 2014, 09:20:02 AM »
A more robust response to Thomas, day one, informing him of GPL infringement would have sent clear signal to all on that thread, prospective purchasers as well, that the GPL licence were being abused.

It doesn't matter whether the GPL license requires a developer to do that or not, it's common sense and removes the chance of pleading ignorance or allowing someone to get away with the infringement quietly without it being public, there's absolutely no reason why a more robust response could not have been made. Clearly setting out GPL infringement.

It can not be excepted for all users on that thread to understand the GPL license, understand why sources we're being requested without making it clear as to why they were being requested and understand how RawMagic infringed GPL, which may have reduced purchases of Rawmagic which also seems to be a bone of contention and no doubt brought the matter to a close far earlier instead of now twelve months on harbouring ill feeling that multiple versions of RawMagic were released with GPL infringement.

ML Developer involvement in that thread aside from requesting sources sent out the wrong signals to ordinary users and Thomas alike. Acceptance of RawMagic by ML Developers by providing help and involvement instead of a brisk robust handling gave Thomas breathing space to continue the infringement and doubt left as to just what action, if any might be taken.

It is surely also the responsibility for the software developers who are having their code infringed upon to assert their position in no uncertain terms as soon as they are aware of an infringement and take robust action, whether that is banning, locking or legal proceedings, not leave it for months / year festering. Rather than hiding behind the terms of the GPL license for months without actually reinforcing them as developers who are having their GPL code violated.

All with the benefit of hindsight, which was my point, there is more that could have been done earlier which wasn't and I say that not as a judgement on any one but to suggest more than one parties inactions prolonged the infringement and have led to where things stand now.

So back to my initial post and this my final, bemoaning Thomas now twelve months on, asking for historic source code and all the other ill feeling is in part a result of in action by ML Developers early on when it could have been nipped in the bud, not one person is solely responsible for the situation now.

2
General Chat / Re: GPL issues with ML post processing software
« on: September 20, 2014, 05:19:56 PM »
Perhaps in the future, you will place more thought into your pathetic observational skills, before making such statements.

 ;)



ouch. :-[

Quote
You're drawing at straws to justify the actions of a singular person (a developer, not a user), who at any time, if he was indeed misinterpreting the request for source code, could have asked a question, "why"!

Ignorance is bliss.

I'm not making judgements, I'm not clutching at straws to try and make a point, its quite simple, I'm saying it's very rare that things are solely the blame of one party, very rarely are things black and white, right and wrong.

What I see here is that there was the chance to nip it in the bud, a robust non abusive, respectful assertion of the facts of GPL on that day in June last year and developers know now that they had the chance, but unfortunately due to inexperience in dealing with such things, as we all learn in life, missed the chance.

The problem with that is ill feeling grows because the individual knows the chance was missed, as a defence they blame the other party, in this case Thomas, like you suggest:

Quote
who at any time, if he was indeed misinterpreting the request for source code, could have asked a question, "why"!

Placing blame solely on Thomas would be naive, it was actions or inactions of all parties that contributed to the mess.

As g3gg0 explained earlier:

A request for sharing the sources, from a ML developer,  in the very first page of the thread, should have been sufficient in my opinion.

That's not really sufficient is it, would you stand up in a court of law in defence of the GPL and think that would have any weight behind it?

I think the lesson to learn is one either acts robustly at the time, make it clear where they take issue or forever try to turn back time blaming one party and trying to claw back some control of the situation. A life lesson that will happen to us all more than once.

Quote
Ignorance is bliss
Yes, and procession is 9/10ths the law and all those other platitudes, fact is if we let people get away with something without being robust at the time, we have nothing.

The answer is not to give them the chance to plead ignorance.

Quote
And my repeating the request before going commercial was very clear.

But where was the advice given regarding GPL infringement in the sequence of events? He's a commercial application developer , do they often provide source code?

I understand your point of view and I'm not wishing to make judgements on whats gone on, just feel some pragmatism is needed and acceptance to learn lessons and move on, not literally of coarse. :-)

3
General Chat / Re: GPL issues with ML post processing software
« on: September 20, 2014, 02:28:49 PM »
So scraxx was the first developer to respond on the thread:

Quote
Good job, can you share sources?

That could be interpreted in many ways. Including wishing to help in development, which would be first on users minds when reading scraxx request, not that scraxx wanted to check for GPL'd code.

What should have happened was establishing GPL compliance, direct question and advising why there was a request for code, if that was the meaning of scraxx's request. Who knows and that's the point no solid response.

Followed by:
Quote
Is it downloadable somewhere? Am I missing something?

Again, open to interpretation. No direct question regarding GPL or advise on GPL infringement.

At this point Thomas hadn't released the app, that came a few posts later, so scraxx request for a download was / would have been assumed by users on that thread, for the app, not the code.

Then Thomas asks:
Quote
Thanks. I'll have a look at the updated source. Can someone tell me what exactly the problem was?

And ML Developer help comes his way #17:
Quote
Here (see comments):

https://bitbucket.org/hudson/magic-lantern/src/tip/modules/lv_rec/raw2dng.c#cl-130

Pointing him to the very GPL code in contention, active developer involvement.

Thomas reply later that day:

Quote
Fixes coming today:

+ 5D Mk II footage now looks right
+ Vertical line problem should be solved (thanks Alex), requires more testing
+ Cancel button

If anyone is experiencing any other issues, let me know ASAP!

There it is. GPL code used in his application, on day one over twelve months ago and no discussion of GPL infringement. The question then is was Thomas aware he was infringing GPL, no one had made him aware of it in fact they help him use it!

So having ill feeling now for no historic code release is something to get over and move on.

This just reinforces the points I made earlier, active ML Developer involvement pushing Rawmagic development onwards without a mention of risks of GPL infringement or straight to the point ultimatum.

ML Developer encouragement continues: #37, same day.

Quote
If you find any situation where the vertical banding is not solved, just upload the first DNG from the video.

At the same time as weakly asking for sources with no reason or advice, active developer involvement with pointing to code and making encouraging remarks, undermines half baked request for code.

And on and on.

4
General Chat / Re: GPL issues with ML post processing software
« on: September 20, 2014, 01:46:03 PM »
@dyfid

The thread hasn't been deleted.

My apologies, I was sure it was stickied, couldn't see it, thought the worst.

Somebody may have a bright idea for a useful app and have the knowledge to implement it on one platform.  If it's good enough, then the idea will be picked up and ported to other platforms.  If not, it will fall by the wayside.  There's no need to unnecessarily stifle creativity.

It's not that hard to write for cross platform, especially a small targeted app for a certain task. Unless there's some specific specialised support required in the OS. Nothing to do with stifling creativity.

5
General Chat / Re: GPL issues with ML post processing software
« on: September 20, 2014, 11:32:23 AM »
I think the Mission Statement and Principles should have two (maybe three) additional entries:

1. ML Developers should refrain from involvement on threads for any applications until it is ascertained if GPL code is used or not and that in the case of yes it is, the necessary actions take place.

This to me is the issue with arguments about RawMagic and particularly ill feeling about previous releases code infringement. It appears a few expressing their ill feeling and arguing points of historic actions, banning and retribution don't appreciate that RawMagic was introduced to ML forums over twelve months ago and ML developer involvement in that thread started at post #17, first page:

http://web.archive.org/web/20130630095753/http://www.magiclantern.fm/forum/index.php?topic=6218.0

ML developer involvement at any level will be seen as encouragement by the application developer and signal acceptance for users to purchase. ML Developer involvement in that RawMagic thread, promoted RawMagic development and promoted sales, luckily it's mac only so....." :-)

The thread was even made a Sticky by forum mods. Then locked, a good move because the History is there to see for users to make up their own minds, then deleted, a pathetic action  which makes users and future users minds up for them.

The lesson has been learnt, you either ACT AT THE TIME and advise the developer, discuss possible GPL infringement, ask for code etc or forever struggle, bitch, harbour ill feeling and gain nothing but learning a lesson, put it down to inexperience and move on. Any move in legal terms would be undermined by the fact ML developers didn't act when they could have and ML developer involvement took place in that thread which will be seen as encouragement. You can thank Thomas for a couple of things, the kick up the ass to tighten up ;-) and the inception of the Mission Statement.

2. Discouraging non cross platform code for open source apps, you know the ones, "I made this app", (even though I ignored perfectly good ones already out there), "it's for" (insert OS of choice) "and just needs this" (insert bullshit runtime environment) "to work".

3. NO JAVA! :-)

6
Post-processing Workflow / Re: Extreme curve comparision
« on: September 17, 2014, 07:23:08 PM »
No doubt frameserving has its uses and so does avisynth but really this is just not it.. Nothing more to say. Bye.

7
Post-processing Workflow / Re: Extreme curve comparision
« on: September 17, 2014, 02:33:15 PM »
But back again to 10 bit. How could I test the AFS's AVI file bit depth?


Running ffmpeg command line should tell you what it's receiving ie: yuv420p etc etc

Quote
In Resolve Lite I can't grade the DNG seqeunce yet, the white balance is not ok and if I start to play with the settings everything will be worse. It hardly can be that this comes from my gappy knowledge. :) Resolve is totally new for me.

Maybe just invest time in learning the basics of Resolve instead of flogging a dead horse. :)

8
Post-processing Workflow / Re: Extreme curve comparision
« on: September 16, 2014, 11:52:14 PM »
With this monster you can playback the graded MLV (DNG sequence + sidecar XMP + AE adjustments) as an AVI file (AFS's AVI) in fullscreen mode, realtime in eg. MPC-HC,

Congratulations and with such a simple workflow. ;) All you need now is MadVR and a display lut, right viewing conditions and your all set up for home cinema quality. ;)

But you've shifted your direction now, no longer talking about 10bit output maintaining quality and you're not getting 10bit still. :)

Do you have sound with your playback? Are you watching an edit? What resolution and duration are your clips off a 550d? Will it work just as well for 5D MKIII?

It's a technical exercise in ignoring the simple route. How many applications you've gone through to get some output, what about tweaking your grade realtime, optical flow speed changes realtime, stabilisation realtime, editing, titling, sync'd sound track.

Resolve Lite will give you all that you've got going on and a lot more besides, in one application.:)

Quote
I want to avoid intermediate filest if I can, we have already the best one, it's the MLV. :) If I can't avoid, yes you have right, but this is a way of experimenting.

Joking apart I get it, it's a technical exercise to play back a DNG sequence with a grade on it but for anyone looking to do more than that it's looks like a right pain in the ass.

Quote
PS: Once I fed MeGUI mod with the AVS file and I saw BGR48 somewhere at the encoding. Unfortunately I saw this only once and I forgot the settings. Anyway interesting thing to experiment with this monster. :D

For the time being.:)

9
Post-processing Workflow / Re: Proress Davinci Resolve Windows???
« on: September 16, 2014, 11:21:30 PM »
For some reason DNxHD makes my footage a little bit more brighter and seemed to give it a green tint as well, also its not very HQ.

But that's nothing to do with DNxHD and everything to do with the media player you're using and how you're encoding the DNxHD, where do you see green tint and brighter output? Which apps? DNxHD is perfectly good intermediate at high bitrates.

The Apple ProRes 4444 12bit is very good and you be hard press to see the difference ,
but  you are limited to Rec709 (HD Color Space 16-235) where the  Cdng's/Tiff's would be linear (0-255) @ 16bit

That's confusing and makes no sense, 12bit Rec709 with levels range 16-235. :) , I know what you mean, limited range but Rec709 is not rgb and it's limited levels range correlates to 0 - 255 rgb, so they are the same.

Quote
Then there is only One,  Apple ProRes 4444 XQ 16bit linear (0-255).

I'm lost I can't find any reference to linear in the link http://support.apple.com/kb/HT5151 where are you getting that from?

10
Post-processing Workflow / Re: Extreme curve comparision
« on: September 16, 2014, 02:46:18 PM »
Avisynth is 8bit, any colour space conversions like rgb <> 4:4:4 using Avisynth functions will be 8bit unless you look into Dither functions and avs2yuv to get interleaved 16bit yuv or rgb, at the moment you're just putting 8bit into 10bit prores / h264 / h265. The whole monster you've created is a total waste of time, honestly pointless. sorry.

Why not do yourself a big favour and just use Resolve?, or as you've already got ACR, Photoshop and AE why not use them? :) Export to a decent intermediate format like DNxHD or image sequences and encode to h264 or h265 from there?

12
Raw Video Postprocessing / Re: Resolve 11.1b color problem
« on: September 15, 2014, 10:27:57 PM »
I'm on a Macbook pro retina 2.3Ghz 16GO RAM Intel Iris Pro so no reason for Resolve to behave like this in my opinion.

It doesn't matter what your opinion is, Resolve is built for NVidia and AMD. Intel graphics to my knowledge are not supported, have you checked minimum specification in the Resolve config guide? What physical screen size?

Quote
And again, not the whole footage is corrupted but only some frames in it, and when I tweak one of those frames (contrast, WB, exposure...) it goes back to normal permanently.

Then it sounds like it's the initial demosaic of the DNG.

Quote
And as I mentioned before, the same footage in lightroom presents no corrupted frame at all.
That's why I deduce it comes from Resolve and not the build, but I might be wrong.

Yes, Resolve is causing the error but it's not a bug in Resolve, your machine does not meet the specification to run it. Workaround is to use a different machine for Resolve, or put up with the iffy performance or stick to LR.

Quote
So no one experiencing the same bugs with the newest beta of Resolve ?

It's not a bug.

13
Raw Video Postprocessing / Re: Resolve 11.1b color problem
« on: September 15, 2014, 09:34:15 AM »
@Levas, with what apps?, all apps, with what demosaic method, all methods? What does dcraw give you demosaic'd and non demosaic'd etc

If it isn't Resolve you're using then wouldn't it be bettet to start a thread?

14
Raw Video Postprocessing / Re: Resolve 11.1b color problem
« on: September 15, 2014, 06:54:39 AM »
Does your graphics card and graphics drivers meet Blackmagics minimum spec for Resolve? What operating system and how are you feeding your monitor?

If it were "simply a problem with Resolve" then all the other Resolve users out there would have same issues. I'd look closer at your setup first.

15
Post-processing Workflow / Re: DaVinci Resolve 10 not reading DNG folders
« on: September 14, 2014, 01:12:47 AM »
Have you checked file / folder permissions for the dngs, have you read the Rawanizer thread and have you used the commandline raw to dng tools to test whether Rawanizer is the problem.

16
Post-processing Workflow / Re: DaVinci Resolve 10 not reading DNG folders
« on: September 13, 2014, 07:53:31 PM »
Then it might be permissions problem with the DNG folders or dngs within.

What software did you use to create the folders and dngs? And what version of Windows.


17
It Be best if you could install Dual GPU's specially for Resolve , as the free version let you use them both.
One as the GUI Monitor and the other for GPU acceleration

Outdated idea to use one for GUI, with earlier versions of Resolve sure, but Resolve 10 & 11 allows using both for compute & GUI together.

Quote
I have a pair of GTX 580 1.5 GB Vram each
In a AMD FX 8350 on a Asus Mother board. I have seen these cards on Ebay used for $150.0 each very good card.
Or the GTX 570 1.2 GB Vram are around $100.0 each used. (have two of those also and very good card)
But for New Cards I would assume you have PICe 2.0 and not PICe3.0 but there are backwards compatible to PICe2.0
I have the GTX 760 2GB Vram , http://www.newegg.com/Product/Product.aspx?Item=N82E16814130932
this is a very strong card lot of cuda cores. The price have come down on that one it's about $250.0 , was close to $380.0 a years ago.
For the dollar to performance level I think that would be a good choice for you.
Or For around $180.0 you can get a GTX660 2GB Vram not a bad card. http://www.newegg.com/Product/Product.aspx?Item=N82E16814133470
but clearly not as good as the GTX 760. I like the GTX 760 for it's have faster ram etc... .

GTX770's are great value, 4GB preferably particularly if you're going to be doing noise reduction in Resolve, more frames can be stored in ram when analysing motion of the noise over a group of frames at a time.

18
It still cant seem to find the DNG folders in the Library even after changing the config settings. only when I add the DNG folder directly to the directory. Is it because i am using a PC or am i just doing something wrong?

Answered on your other thread.

http://www.magiclantern.fm/forum/index.php?topic=13282.msg128280#new

19
Post-processing Workflow / Re: DaVinci Resolve 10 not reading DNG folders
« on: September 13, 2014, 09:51:52 AM »
When you say configure what do you mean? Have you gone into the Resolve Menu and under Preferences added the folder to Scratch Disks? Other than that maybe a folder permissions problem.

btw It's not a problem detecting DNG's or folders with Resolve 9, 10 or 11. Really do upgrade to 11 though as it has more advanced CinemaDNG raw adjustments as you're probably already aware.

20
Post-processing Workflow / Re: Always Underexposed :( Any help?
« on: September 06, 2014, 02:54:35 PM »
Dyfid,

Thank you contributing your time and effort into helping me so much, i really appreaciate this (the guys over at EOSHD dont even bother helping you)

That's kind of you to say but am I really helping. :)

Quote
Reason why i am using BMD Film is beceause i wanna get rid of the ''video'' look

But what do you mean by video look? Is that really your goal?  Applying a film emulation lut, some recorded film grain loop and a logish gamma curve is not going to imho remove the video look if the underlying shots motion isn't right, so much is to do with motion, framing, movement through the frame, transition from one scene to the next, how one scene relates to another, what you're trying to say, the feeling and emotion you're trying to evoke. Otherwise you could end up with a video that looks like it was shot on an mobile phone obfuscated with layers of cr*p to try and make it look filmic.

Quote
and try to emulate a film stock, Kodak's legendary Vision3 500T 5219 to be exact,

Sure and that's cool, we do what we like, it's supposed to be fun right. :) I personally couldn't tell one film stock from another. :)

Quote
and rec709 simply is to ''baked'' for me.

Except it's not in the case of raw, you can adjust under the curve?

Quote
so to my understanding i really only should worry about getting it right IN camera since i can get that information back after

Using the ML raw exposure tools to avoid to much raw channel clipping will give you the most out of the camera. Clipping a raw channel too much will be detrimental to what you can get out of the raw data, even if it is raw because ultimately when applying WB to get a neutral starting point, the remaining channels levels will be scaled to the one that is most exposed depending on the light and colour in the scene, possibly dragging up noise in the weaker channels.

Quote
BMD Film because it simply compresses the highlights and shadows right?

Imho the purpose of log is at capture time, to compress high performing cameras with wide DR into a economic format that maximises capture in minimal acceptable bit depth and compromised codec at the time of capture. When using linear would demand high bit depth storage and result in excessive file sizes making excessive impact on capture media and therefore cost, management and handling.

And that applying a BMD log curve to 16bit raw linear data when working end to end in the same application such as Resolve 11 appears to be pointless and nothing more than a one shot 'fix all' before the lut f**kery. :)



21
Post-processing Workflow / Re: Always Underexposed :( Any help?
« on: September 06, 2014, 10:33:28 AM »
Going back to your original post and rereading your issue is with BMD color space applied to ML raw, as Levas says BMD on ML raw will give the appearance of underexposure whether you use the raw exposure tools or not.

But with the raw tools, exposure is more accurate from the point of view that we see if clipping is occurring in the raw data which ultimately will affect how well it can be post white balanced, how much DR is captured after WB and demosaic and whether or not as a last resort highlight recovery of some sort has to be used to try and remedy bad exposure control from the point of view of raw data clipping. Blindly applying HR can screw up badly exposed shots even more than help.

But going back to BMD as a starting point for lut f**kery, I understand that's what Visoncolor recommend, but with Resolve 11's camera raw adjustments that happen under the rec709 curve and gamut is there really any need to push your raw data into a logish curve like BMD, personally I don't think so. It's 16bit linear raw data filtered through a rec709 curve, which at that bit depth will provide more than enough levels to work within and you are in under the curve to adjust raw data before going to more typical 1D tools like RGB curves and LGG, even with the rec709 2.2 ish gamma applied without resorting to log, if going end to end in Resolve 11.

But you have the Repulsz luts so your own tests will help you decide, would be interesting to hear your findings.


22
Post-processing Workflow / Re: Always Underexposed :( Any help?
« on: September 05, 2014, 02:19:33 PM »
Try using the raw exposure features as Audionut suggested.  ;) The raw exposure tools aren't affected by picture style. Then just use a picture style to help with focus if using manual lenses, such as Landscape or maybe like me you swap between h264 & raw, so Landscape for viewing, VisionTech for recording (h264).

I have no problems with ML raw in Resolve, when using the raw exposure tools in ML ;), the raw interpretation looks just fine with rec709, no underexposure unless I was struggling with light.

Using BMD Film is pointless imho for ML raw same as Highlight Recovery which is unnecessary if exposing 'accurately' using the raw exposure tools in ML ;), ie: minor clipping in spec highlights.

If memory serves me right ML raw recording was a discovery by the devs whilst creating the raw exposure tools ;) in preference to UniWB.

23
It's a commercial product with a time limited beta release? Oct 2014?

OS really doesn't matter as I'd assume it would be run on a dedicated machine, to get realtime playback of multiple scopes, use a seperate monitor and GPU assisted using a feed from your grading app free of any OS level colour management such as through a BM mini monitor.

Alternatives would be something like BM Ultrascope.

24
Raw Video / Re: 700D / T5i RAW video
« on: August 26, 2014, 11:33:42 PM »
To remove the auto focus pattern of dots use chroma smoothing, like CS 2x2, in whatever you're using going RAW or MLV to DNG. Alternative is pink dot remover (PDR) I think but never tried it.

Like you I'm using a 700d and 550d. I'm using Sandisk 95mb/s for the 700d and Sandisk 45mb/s for 550d. Recent builds with SRM module which doubles record time, making the 550d more usable for raw again and doesn't have the focus screen dots to remove.

25
Handbrake is cross platform, open source, Windows, mac & Linux. x264 is also cross platform and open source. I think there's some x264 plugin type things about for Premiere, not really looked into that though.

Pages: [1] 2 3