I tell you what...I invite those of you arguing to go out and test it for yourselves to prove or disprove this theory. @nothing...don't bring the arguments over here please. It's really because of you that the thread was shut down. If the guy will not budge on what YOU want then maybe it's just time for YOU to go out and prove he's wrong. Don't turn this into another he-said-she-said kind of argument. If you just don't want to try it yourself and chose not to believe his argument...then just walk away. Honestly Aki isn't bringing up anything that I have not heard of before and I understand what dikitsov is trying to say earlier...but...prove it right...prove it wrong or...just stop guessing/arguing.

@Aki...my suggestion to you is this: There have been many requests from folks here asking for you to do a before (Raw/AVCHD file from camera) and after with video. Before responding to any other threads about the benefits of this or that....just produce what the readers want to see. I understand you want to demonstrate your workflow.....but let that be last on the priority list. I think some just don't care about workflow until they see "actual" results from a before and after video. It's at that point where a workflow will make sense (or not) for everybody.

EDIT: @nothing and @zcream FYI...it's not that I don't agree with you btw.

I'd still love to see some comparison footage that shows the claimed improvement, even if Aki threatens to sue me again.

I feel like banging my head over the arguments on that thread. Somewhat glad that kholi shut down the thread. AAAARRRGGHHH

So the choices are :- Correct WB for daylight and well lit scenes. For low light=

1. use 5000K WB

2. Alternatively use a 85b or magenta filter and accept 2/3 stop light reduction.

3. WB set at correct color temp (or 3200K) pushed about some clicks to blue and some down towards magenta.

Is this a reasonable summary ?

What does this mean ?

Очень низкая чувствительность красного канала относительно зеленого (баланс белого на 1.6 стопа)

@aki_hartikainen From what I gather, your recent color filter tests were all done on the GH1. In my experience, that camera has a greenish cast, compared to the GH2. When shooting footage on the GH1, I correct for this by setting the GH1's WB+ adjustment to M-5 (after setting the white balance for the scene). That seems to do the trick:

In the video above, the first clip is from the GH1, the second from the GH2, and the third from the Nikon D5100. Color correction was done on the GH1 in-camera.

On both GH1 and GH2, there's an easy way to check out the individual R, G, and B histograms before you shoot. Make sure the camera is set to sRGB mode, and that Manual still photo mode is set the same as Creative Movie Mode (in Manual exposure mode of course). Then frame and expose your shot in Creative Movie Mode, and switch to Manual photo mode to snap a still frame. (This insures that your still photo settings match your video settings.)

In Playback mode, you can then display the individual R, G, B, and Y histograms next to the photo and compare their levels. When I do this with a broad-spectrum scene using my M-5 setting, all three color channels turn out reasonably well-balanced, and require no further color correction.

All the theory talk means nothing if the final results prove to be an improvement! Theoretically Bumble Bees shouldn't be able to fly... If the guy is finding visual improvements using his filters, then perhaps there's something happening that the sensor theory isn't taking into account.

@CRFilms @dkitsov ...yes never shoot with ND filters outdoors, just use shutter and apature to get the best image. >_> Logical fallacy. ND filter ideally should not change the color balance neither should the aperture/shutter speed combination. This discussion is about changing the color of the light that reaches the sensor sights, not about the means of overall exposure control. Sensor needs as much light as it needs. Ways of limiting it with ND or shutter/aperture, while makes aesthetic, difference probably makes no "electrical" difference as far as ISO adjustment is not involved (that would move the sensor from the optimum operating range)

It depends completely on the lighting situation. If it's daylight with very high energy in blue – in elevated places at noon it can be much more than 6500 K – you'll blow the blue channel very early with purple (or any filters other than orange).

IMHO, any colored filter only makes sense to decrease the strongest channel to avoid starving the others if there is a massive misbalance. It doesn't really help much with the 4:2:0 vs. 4:2:2 issue.

@dkitsov, I thought the idea was that you're allowing the sensor's Blue and Red channel to get more saturation by basically pushing the amount of light in those bands to hit the sensor as opposed to the dominant Green, which will always have more sensitivity than the other channels. So Blue and Red get what they need in order to lower noise in the Blue and Red. I may not be describing it well, but that's how I understood it.

http://www.libraw.org/articles/magenta-filters-on-digicam.html

"The point in using magenta filter is to achieve better exposure of red and blue channels by allowing less light to hit the sensors under the green filters of sensor CFA. This is physical, pre-capture filtration; and as such it can't be reproduced by tuning the camera or through post-processing."

The guy just found this article from 2008 talking about how a magenta filter can create a better image in your DSLR shooting in daylight. Said there is a sweet spot for WB with a filter that can give less noise, but it's different for each hack and ISO so you'll have to test the settings yourself to figure out what works for you at what ISO, bitrate and WB.

I got my filters and did a quick shoot outside and the purple filter gives the outdoor footage a nice look. File Purple is the image with the filter, PurpCC is the video with some basic color correcting.

I'm very sure that there is no difference between setting your WB to pink/magenta and using a colour filter, (unless you're after some kind of vignetting 8 or 16mm film effect and you've got a graduated filter for the job, yeah, good look CCing that to a baseline). I just did some quick tests. All shots were in the Smooth profile all at -2. In order there is a shot of the scene with normal WB, then one where I set WB to green to get pink. The next shot is the version of the pink WB I corrected (not perfect but in the rough ballpark really), then there are shots of the red channel of both the correct WB and the corrected pink WB shots.

The red channel is different, I'll give it that. There does seem to be much better definition of the edges in the shot, smoother shifts (less chance of banding), and better response in the shadows. (more detail perhaps ?). Correcting the footage does make for deeper, and in this case, more accurate reds. Anyhow, more, proper tests are needed. As you can see it creates quite a big difference with only a subtle pink WB shift (although you do drop a fair bit of the overall light too). I'm not sure if this is giving more detail/better chroma sampling, or if its just causing a colour shift to red after correction, although the evidence in the red channel does indicate something positive is happening.

All shots were made with the @bkmcwd GOLGOP3 patch with the Sedna matrix with a GH2. Oh, another note, the average bitrate did drop from 130Mbps to 115Mbps from the correct WB to the pink WB.

Edit : Added a better correction of the pink WB version.