IMO it's suspicious that nothing joined personal-view just after that thread was closed. Just to continue trolling @ PV?

@SightfulProd that's what I like about at P-V. Here isn't really any fights. At one Finnish video forum you cannot even say word "GH2" without fight, because "oldskool professionals" don't see anything good at it.

I posted a couple of times in the other thread, and here in this one when I made a quick attempt to see if the method did improve the sampling in the red/blue channels. For me the pain of correction far outweighs any benefit (a benefit which in itself is still inconclusive). I concur that Aki did make some pretty strong claims about curing posterisation with this approach. For me it was never about that, it was about does it actually improve sampling in the red/blue channels. Now, that could arguably improve posterisation, but I'd never expect it to result in it never occuring again. It would though give us more latitude when grading/colour correcting if we had better sampled red/blue channels.

Aki though went on to create and argue for a very specifc workflow, involving specific hacks, specific lenses, specific filters and specific software. A workflow which a lot of us know is unnecessary (and horribly painful when you consider it involves working with stills in batch with silkypix). He also argued for saving our footage as 16 bit Tiff files in order to benefit fully from silkypixs 'virtual raw'. All virtual raw is is the same kind of upsampling that pretty much all of our NLEs, colour correction and compositing tools do. We know that of itself that process does not magically improve our source footage. Sure, interpolation can be done when we upsample from 4:2:0 to 4:2:2 or higher which does improve the channels, and I'd argue that this produces results as good as the benefit of the filter approach. In fact, a good denoising tool does this also. I've noticed a remarkably improvement in the blue/red channels using the wavelet based denoiser in Nuke for example. (edit for clarity: But that is never going to cure banding/posterisation thats already there). There may be arguments for working with, exporting, 16 bit files in a compositing situation and those of us who need to do that know when we need to do it. But frankly, for most situations 12bit DPX is probably overkill when your source is AVCHD (though 10bit can make sense I think), no matter how much you've managed to improve the channels during or after upsampling.

Aki never managed to show an example where he had truly, completely, removed the filters cast from the source footage. Nor have I for that matter, the more I thought about it, the more I Thought that the benefit to pain ratio made it not worth it, and so I decided to stop investigating it.

I can see that it is worthwhile to investigate the positive effect the filter could have on the quality of 4:2:0 source, but everything else Aki is suggesting, to me, makes very little sense and some if it just sounds like snake oil.

Auto WB with filter http://www.mediafire.com/?bg65c8t9b2io556

Auto WB no filter http://www.mediafire.com/?4w5nge1g3vsnw5l

3500k no filter http://www.mediafire.com/?u89vuhu8ioe0gsu

3500k with filter and green n blue sliders moved to top right.

http://www.mediafire.com/?vqw5k6u6c2xfc9k

3500k with filter no other WB adjustment.

http://www.mediafire.com/?5pz5fhu0scd38cp

Shot with 14mm f2.5 panny at 2.5 and various iso and shutter to try and match exposure. Max iso is 400, low iso is 320 and max shutter is around 160, low shutter is 50-60. Can't remember exactly it says on the videos. I also misspoke that WB was 3200k in some of the videos, but it was set at 3500k.

This is a cheap purple cokin style plastic filter so it softens the image a tad vs stock. I only have Vegas Studio not Pro or any color correcting software so those of you with pro NLEs please take a look and tell if you think the red and blue channels look better with filter.

And I agree the whole "virtual raw" workflow is way too much hassle, but outdoors I really like how the filtered footage looks. Just need a proper optical glass filter and to know, EXACTLY how much purple is needed for best results.

FINAL EDIT: Cam used GH1, 14mm Panny, Max Latitude hack.

@kholi "I initially thought that people were being a little vicious with Aki, but shut the thread down for other reasons."

You also chose not to communicate with me at all regarding moving the thread or removing some offending remarks, so what were those other reasons? Several people were already testing and using the methods besides myself, and they hoped that the thread could have continued in a less heated environment.

No I havent tried it, because I know its not necessary, and silkypix is offering nothing over other applications that would make a lot more sense to use. Its also extremely painful, just as one example, can you scrub through your frames to see how your corrections hold up in silkypix ? I'm not interested in continuing this discussion until you show that you actually understand a normal grading/CC workflow, and manage to technically argue rationally how your workflow is better than any other approach. Granted, as I admitted in the other thread 'virtual raw' is silkypixs terminology not yours, but its no different to what any other NLE, colour correction or compositing application does with source material. You say a lot of things such as 'software tailored for the camera' and don't ever back them up with a proper technical explanation (neither does the silkypix documentation). So, thats why some people like myself don't take you seriously. If you had just stuck to testing the efficacy of the filter on the red/blue channels after correction you wouldn't have had so many people challenging your ideas. Are you not suggesting specific lenses now ? Or specific hacks ? I use a GH2 so I can't play along with the specific hacks, and I've no intention of buying lenses to test your workflow either.

Saying that there is no need for denoisers or deblockers at all, is just ridiculous and shows a lack of experience to me. Your demo still has a purple cast after correction, and I see no additional detail or lack of noise in the elevated blacks over other footage shot normally. Again, as others have requested, shoot proper comparisons in controlled conditions. Also, compare at least one other accepted workflow to yours and post the results of both. How do you expect any of us to take it seriously if you haven't done, and provided, these basic comparative tests (as your shot of the guy in the orange hoodie proves, posting stills is not enough). Until you can completely remove the purple cast any comparisons are pointless as you've not managed to achieve a baseline for comparison. Your opinions are your opinions, if people agree with you and have the time to follow your workflow, and believe in your workflow, then fair enough. Just accept that I won't.

I didn't use the term snake oil to mean you're selling something (as in cash money), but you definitely seem to be selling an idea with statements that have no real technical basis.

CRFilms, thank you for posting this comparison. This is the kind of thing I've been asking for. Where do you see an improvement?

@stray "You say a lot of things such as 'software tailored for the camera' and don't ever back them up with a proper technical explanation (neither does the silkypix documentation). So, thats why some people like myself don't take you seriously."

Says on the Panasonic Silkypix front page: "Development and Adjustment of JPEG / TIFF Files

This utility converts data into a form that is close to RAW data and then re-applies the development processing. This ensures accurate color processing.*"

http://panasonic.net/avc/lumix/silkypix/index.html

To do that would mean that the software has parameters to know how a jpeg taken with this camera should look like, had it been taken as RAW. It requires a software spesific to the camera and not some random generic software.

@CRFilms thanks for the good test files, I've had a quick play with the Auto WB files. The maximum and minimum luminance of both shots, with and without the filter, are pretty much identical. Anyway in Nuke I used a colour lookup node based solely on remapping the average intensity of both shot, and then I've used the match grade plugin. I'm trying to keep it as simple as possible here. Still not got a perfect correction though (check the shadows, the bottle reflection/diffraction and the wood), but this is as good as I can get without doing an awful lot of remapping and further correction. I think though its good enough to decide if you think this approach is worthwhile.

While there does on first glance seem to be a noticeable benefit to the blue channel its actually not that good, in fact the blue channel isn't 'right', check the text on the cereal. Therefore you still have to say that what I'm posting is still inconclusive. In terms of the lack of blockiness/noise then yep, I think thats the case, but good denoising would sort that and would also result in smoother, interpolated red & blue channels anyway. I'll go further and say its actually having a negative effect, it isn't improving the sampling of the blue or red channel at all in terms of retaining colour information (not noise, noise does seem to be reduced), quite the contrary in fact. As for if it helps with posterisation/banding you can't tell from this shot, but I'm doubting that too now. I think what @lpowell and @driftwood are doing with changes to the encoders matrix is having more of a positive effect to how colour is encoded than this approach is (until we get to alter film modes).

There is an imperceptible flicker caused by the light in this shot, so the average intensity is changing. This means that whatever solution you use to remove the filters cast is probably going to have to be adaptive. This will be true of shots in natural light too. As I and others said, to do this properly you are going to have to use a LUT for the correction based on good filtered/unfiltered shots of a colour chart in studio conditions where you are absolutely sure that you are covering the cameras entire dynamic range. Even then thats not only going to be specific to the filter, but to the lens used and the film mode. I also think that although it would give much better results in terms of removing the filters cast it still won't be perfect. As we all have enough issues getting accurate colours as it is, and this is just adding to that issue I can't see any point in it. Sure I could maybe just stop correcting out the filter now and go for the grade I want, but I'll be dealing with additional, (and to me anyway, counter-intuitive) colour problems.

Edit : Forgot to say, yes, the image appears to be softer, maybe thats the filter. Or it could be the matchgrade node, I really distrust that thing.

In the files that I compared above the filtered version contains 3.6% more data despite being slightly softer. I don't see what else that could be other than the beefier colors. The filter modifies the contrast both indoors and outdoors.

I've known about this all along, because I kept monitoring the bit rates. Bit rates fell when I removed the filter and went up with the filter. I tuned the current 25p iW settings so that they remain on the edge. Without filter the settings record in all situations, with filter I get occasional write errors. I am going to publish my own measurements later, there is more to do right now than just to win an Internet argument.

@aki_hartikainen - "I think lpowell does that too, he was rather confused about it in the deleted thread."

I think it's quite clear where the original source of confusion lies.

Aki. A 3% variation on a single sample file is statistically insignificant. Especially since a minor change in the scenery can account for this discrepancy. It does not look that anything significant is available. Also lay off @lpowell

@aki_hartikainen

Please try not to lead this topic into conflicts.

I am really sick of al this claims, demands, etc.