Teodor Miljevic
Active member
Dear friends even though I don't write here often I frequent this forum almost everyday. I've been following The AF 100 forum And I must say that I am a little underwhelmed with the look of the camera especially in the highlights. Before people from both sides of this argument start to bombard this thread please do me a favour and do not do this. There are several subjective threads already discussing this never ending discussion based on taste (hence it will never be resolved). So no matter if you are a fan of the look or a little disapointed (as I) I would like to have a little discussion about maximising it's MOJO with some external addons namely fiters.
First a little background. Before I shoot a independent feature called Weekend on the dvx 100A. I stumbled over an article talking about filters to complement the then popular cameras such as the dvx100a pd150 and xl1. I read that the tiffen golddiffusion fx and black diffusion fx filters (both 1/2 strength) helped and after testing them I could conclude that the picture actually became a little nicer with ever so subtle highligt rolloff and a nice boost in the reeds (gold diffusion)
The reason I'm mentioning this is that I would like to start a thread on using filters with AF100 to what I and many of those who agree with me is the cameras biggest problem namely highlight handling.
Why do I propose filters? Because the AF100 is 8 bit I think you need to bake as much of the look as possible in to the finished picture. I also think that the 8 -bit is the sollution as to why the AF100 handles highlights poorly (this is a subjective notion so if you don't agree with me thats fine then you can be motivated to contribute(in the form of tests with filters) to make the AF100's highlights even more pleasing if that is your opinion) It was a great problem for panasonic (I'm speculating) to not make the cinema D setting to flat since it would canibalize on the midtones (even though I think they should make an even flatter setting for high contrast situation avaliable through firmware update). The optimization is also evident if you check all material that is filmed in overcast situations (where it produces a very pleasing picture from scratch which I find points a little to some of the mistweakings in the cinema D setting since I think a overcast day should produce a more lowcontrast image (but I also understand since you have to spread that picture over those 8 bits).
So I find two areas where filters might help the first one is with the slightly to contrasty Cinema D (which should be the flattest setting that the AF100 has if I'm not mistaken) which results in harsh clipped highlites. The second problem is the the ugly miscolorations in the blown highlites. Here is a question If the highlite is discolored is it the the channel with the same color as the discoloration that is maxes out or is it the oposite (the other channels have maxed out. Since I dont know the answer my theory about the second problem will be theoretical and depending on the answer to the question posed above).
Before anyone states the obvious I would like to point out that I know that there are problems with using filters with loss of resolution, flaring and unforseen brightening of the picture (during movement with lowcon filters).
The problems that I try to adress are regarding high contrast situations since the AF 100 looks quite nice in low contrast.
Since I don't have a AF100 (and probably won't have if I don't get an indication through this discussion or some other nice persons hard work that high contrast can be handeld with more grace.) I need your help and as I said I think a methodic testing in the lines of what I'm about to propose is benefitial for fans as well as critics.
For the first "problem" I would propose some kind of light diffusion filter such as the golddiffusion fx black diffusion fx or difussion fx hd from tiffen I would try 1/8 1/4 and 1/2 in strength there are othe manufacturers out there and you can try them as well I just don't know them so thats why I'm not mentioning them. If you want to keep the picture lower contrast (and a little bit softer) try to avoid softfilters that flare. Black diffusion tends to flair less in the highlites then regular difusion. Another thing worth testing is low contrast filters. There are diffierrent flavours but here I would recomend testing lower grades 1 and less to see if it is possible to find something that can be used more consistently in High contrast and not just under special circumstances as a locked down camera.
The second "problem" is about fake colors in highlites. This is due to in theory as I understand its channels maxing out at different levels so some channels are maxed out while others still retain some detail. As I stated in my question above I don't remenber which maxed out channels that gives what discoloration. But if we for the sake of argument say that it is based on oposite channels (if I'm wrong the only difference would be that we handle the problem with the inverted colors (or oposite (in the color wheel) to what I propose here)) Then the yellow discloration seen in many blown out highlites is due to a maxed out blue channel. I have seen the following fake colors in highlites mostly yellow and then even some magenta. if it's as I said that the opposite maxed out channel gives the opposite fakecolor that would mean that on the AF100 the blue channel is the most sensitive then comes the green and last comes the red (Since magenta fakecolor comes from green beeing maxed out)
So I'm proposing that if we tone down the blue and the green element in the picture that our highlites would be less prone to fake colors. Here i would also like to propose that we split the testing into daylight and tungsten since the whitebalance can have an effect differently depending on what whitebalance we have.
How do we do this? In theory it is easy to at least try. You need a colored filter in the same color as the discoloration since that acts as a cutfilter for the other colors. If you have a yellow filter it is like a gateway that lets through more yellow (red and green) then blue so it would in effect reduce the blue channel. Here I think we need testing with both whitebalance before filter and a session with whitebalance after filter. The two different methods test different thing that could be equally interesting for making conclusion.
whitebalance before filtering.
This could tell us if there is indeed a unbalance in the highlite handling. If we whitebalande and then point our camera towards a problematic area with false colors for example yellow. We can then keep that balance and apply a yellow filter (this would make the picture more warm but it would also cut the blue channel) if this lessens the problem and the highlites appear more neutral the there is a blue bias in the highlites.
If we do the whitebalance after the filter that means that we have cut the blue frequencies before they hit the chip and then we ask the camera to compensate meaning that the blue channel will be boosted. If the problem are even worse with false color (then pointing the camera and doing a regular whitebalance without filter) then the electronics that are involved are the culprit then a method of putting on a lightblue(very light) filter to cut down the red and green frequencies can be a good method of reducing the electronics involvement (not having to boost yhe blue channel as much.
There are several problems with a fix like this such as lightloss and having to record squeved color (if the whitebalance before filter works) and added noise. Probably such a fix would be best handled if you record to a better codec in 4:2:2 so you can post the colors back.
How do we test this easily, well. To test the theory we could use lightingfilters such as ctb, and straw for yellow (actually since green acts up as well sometimes I think the ctb in different grades would be ideal as the "yellow" filter. Test with grades ranging from 1/8 up to 1/2. To boost the blue channel test ctb in strengths varying from 1/8 to 1/2. If you have a mattebox make sure that you cut a piece that is large enough to cover the front. This is not a permanent fix so don't mind some glare from the filter just focus on cutting the channels so we can see how it influences the highlites.
I think tests like this can lead to a few different things. Firstly we can get an external fix or reduction of some of the problems we have spotted with the Af100. Secondly if we go to the bottom with the color channels maybe panasonic can inplement some of our findings in a firmware update (tweaking the gain and the curves for the different channels and their highlight handling and maybe giving us a more flat Cinema D called cinema DD
and the third maybe some great company makes a MOJO filter that softens the picure a bit, lessens the contrast a tad and reduces the proper color channels to reduce highlite problems, maybe they can be called AF 100 MOJO FX DAY and AF 100 MOJO FX Tungsten
Please I beg you don't make this thread in to I like or I dislike the AF100. Let us use our brains and delivered cameras to find out how we can maximize this little baby
First a little background. Before I shoot a independent feature called Weekend on the dvx 100A. I stumbled over an article talking about filters to complement the then popular cameras such as the dvx100a pd150 and xl1. I read that the tiffen golddiffusion fx and black diffusion fx filters (both 1/2 strength) helped and after testing them I could conclude that the picture actually became a little nicer with ever so subtle highligt rolloff and a nice boost in the reeds (gold diffusion)
The reason I'm mentioning this is that I would like to start a thread on using filters with AF100 to what I and many of those who agree with me is the cameras biggest problem namely highlight handling.
Why do I propose filters? Because the AF100 is 8 bit I think you need to bake as much of the look as possible in to the finished picture. I also think that the 8 -bit is the sollution as to why the AF100 handles highlights poorly (this is a subjective notion so if you don't agree with me thats fine then you can be motivated to contribute(in the form of tests with filters) to make the AF100's highlights even more pleasing if that is your opinion) It was a great problem for panasonic (I'm speculating) to not make the cinema D setting to flat since it would canibalize on the midtones (even though I think they should make an even flatter setting for high contrast situation avaliable through firmware update). The optimization is also evident if you check all material that is filmed in overcast situations (where it produces a very pleasing picture from scratch which I find points a little to some of the mistweakings in the cinema D setting since I think a overcast day should produce a more lowcontrast image (but I also understand since you have to spread that picture over those 8 bits).
So I find two areas where filters might help the first one is with the slightly to contrasty Cinema D (which should be the flattest setting that the AF100 has if I'm not mistaken) which results in harsh clipped highlites. The second problem is the the ugly miscolorations in the blown highlites. Here is a question If the highlite is discolored is it the the channel with the same color as the discoloration that is maxes out or is it the oposite (the other channels have maxed out. Since I dont know the answer my theory about the second problem will be theoretical and depending on the answer to the question posed above).
Before anyone states the obvious I would like to point out that I know that there are problems with using filters with loss of resolution, flaring and unforseen brightening of the picture (during movement with lowcon filters).
The problems that I try to adress are regarding high contrast situations since the AF 100 looks quite nice in low contrast.
Since I don't have a AF100 (and probably won't have if I don't get an indication through this discussion or some other nice persons hard work that high contrast can be handeld with more grace.) I need your help and as I said I think a methodic testing in the lines of what I'm about to propose is benefitial for fans as well as critics.
For the first "problem" I would propose some kind of light diffusion filter such as the golddiffusion fx black diffusion fx or difussion fx hd from tiffen I would try 1/8 1/4 and 1/2 in strength there are othe manufacturers out there and you can try them as well I just don't know them so thats why I'm not mentioning them. If you want to keep the picture lower contrast (and a little bit softer) try to avoid softfilters that flare. Black diffusion tends to flair less in the highlites then regular difusion. Another thing worth testing is low contrast filters. There are diffierrent flavours but here I would recomend testing lower grades 1 and less to see if it is possible to find something that can be used more consistently in High contrast and not just under special circumstances as a locked down camera.
The second "problem" is about fake colors in highlites. This is due to in theory as I understand its channels maxing out at different levels so some channels are maxed out while others still retain some detail. As I stated in my question above I don't remenber which maxed out channels that gives what discoloration. But if we for the sake of argument say that it is based on oposite channels (if I'm wrong the only difference would be that we handle the problem with the inverted colors (or oposite (in the color wheel) to what I propose here)) Then the yellow discloration seen in many blown out highlites is due to a maxed out blue channel. I have seen the following fake colors in highlites mostly yellow and then even some magenta. if it's as I said that the opposite maxed out channel gives the opposite fakecolor that would mean that on the AF100 the blue channel is the most sensitive then comes the green and last comes the red (Since magenta fakecolor comes from green beeing maxed out)
So I'm proposing that if we tone down the blue and the green element in the picture that our highlites would be less prone to fake colors. Here i would also like to propose that we split the testing into daylight and tungsten since the whitebalance can have an effect differently depending on what whitebalance we have.
How do we do this? In theory it is easy to at least try. You need a colored filter in the same color as the discoloration since that acts as a cutfilter for the other colors. If you have a yellow filter it is like a gateway that lets through more yellow (red and green) then blue so it would in effect reduce the blue channel. Here I think we need testing with both whitebalance before filter and a session with whitebalance after filter. The two different methods test different thing that could be equally interesting for making conclusion.
whitebalance before filtering.
This could tell us if there is indeed a unbalance in the highlite handling. If we whitebalande and then point our camera towards a problematic area with false colors for example yellow. We can then keep that balance and apply a yellow filter (this would make the picture more warm but it would also cut the blue channel) if this lessens the problem and the highlites appear more neutral the there is a blue bias in the highlites.
If we do the whitebalance after the filter that means that we have cut the blue frequencies before they hit the chip and then we ask the camera to compensate meaning that the blue channel will be boosted. If the problem are even worse with false color (then pointing the camera and doing a regular whitebalance without filter) then the electronics that are involved are the culprit then a method of putting on a lightblue(very light) filter to cut down the red and green frequencies can be a good method of reducing the electronics involvement (not having to boost yhe blue channel as much.
There are several problems with a fix like this such as lightloss and having to record squeved color (if the whitebalance before filter works) and added noise. Probably such a fix would be best handled if you record to a better codec in 4:2:2 so you can post the colors back.
How do we test this easily, well. To test the theory we could use lightingfilters such as ctb, and straw for yellow (actually since green acts up as well sometimes I think the ctb in different grades would be ideal as the "yellow" filter. Test with grades ranging from 1/8 up to 1/2. To boost the blue channel test ctb in strengths varying from 1/8 to 1/2. If you have a mattebox make sure that you cut a piece that is large enough to cover the front. This is not a permanent fix so don't mind some glare from the filter just focus on cutting the channels so we can see how it influences the highlites.
I think tests like this can lead to a few different things. Firstly we can get an external fix or reduction of some of the problems we have spotted with the Af100. Secondly if we go to the bottom with the color channels maybe panasonic can inplement some of our findings in a firmware update (tweaking the gain and the curves for the different channels and their highlight handling and maybe giving us a more flat Cinema D called cinema DD
and the third maybe some great company makes a MOJO filter that softens the picure a bit, lessens the contrast a tad and reduces the proper color channels to reduce highlite problems, maybe they can be called AF 100 MOJO FX DAY and AF 100 MOJO FX Tungsten Please I beg you don't make this thread in to I like or I dislike the AF100. Let us use our brains and delivered cameras to find out how we can maximize this little baby
