HI Terry

As I see it the statement -

Is the implication that these controls in MEP have an inadequate range, or what? If I can get good results from my 10 year old image editor I'd expect MEP to be similarly capable.

- is that you are trying to  compare two different ways of building up a picture. In simple terms: -

A still image has all the colour information needed and no movement to cater for and can be adjusted as a whole.

A video is not a series of complete still images with full image information projected at 25 (30) fps to create the illusion of movement.

A video consists of I, P and B images (frames) which are (simplifying here) -

I    -  image is a full image,

P  - is a predicted image with only the changes from the previous I or P image

B  - is a bidirectional predicted image which contains the predicted differences between the previous / next image to the next/previous image - ie the codec looks ahead / behind to see what has changed and builds an in between image.

A video may therefore have the images arranged like this (theoretically)

          I P P B I P P B I    .......................

As you can see in this section of a theoretical video only 25%  of the 'images are actually a 'full' image - the others are the calculated differences and only contain image information that changes, the less the movement between images the less information in the P and B images.

Consequently any adjustments that you make have to take these calculated images in to account and the amount of adjustment you have is restricted because the P and B images are only partial images.

In reality it is even more complicated then this,because the number of P and B images between the  I images can vary.

Hope this makes sense

John

Hi Peter,

Yes thanks, I tried Color too but still couldn't get a successful result. Even when I chose a large area covering the various shades of green (including the dark creases in my un-ironed and poorly lit sceen).

Can't say I noticed the quirk you mentioned, but I'll try it again this morning.

Hi John,

Have to admit I didn't really follow that! Maybe after a coffee or two I'll try again.

You say "A video is not a series of complete still images with full image information projected at 25 (30) fps to create the illusion of movement." But that's just what I understand a digital movie to be, in effect, namely a series of frames. Surely, however they're constructed (and I know that's complex), that's what we end up seeing.

If I use Ctrl + 1 in MEP I can study each single frame. If I exported every frame and painstakingly used PSP8 to convert its green background to transparency, then I could re-import them, add a new background (surface of the moon, for instance) and play my intended movie.

So my hangup is that although PSP8 (and I've no doubt Photoshop and other image editors) can remove uneven green backgrounds, MEP doesn't seem able to do so. And I still don't see why, apart from my guess that its RGB selection tools are somehow inferior.

--------------------

Here's are some self-explanatory illustrations:

--
Terry, East Grinstead, UK

Hi Terry

I agree the whole thing a a mind twister.

You say "A video is not a series of complete still images with full image information projected at 25 (30) fps to create the illusion of movement." But that's just what I understand a digital movie to be, in effect, namely a series of frames. Surely, however they're constructed (and I know that's complex), that's what we end up seeing.

What you end up seeing is an illusion of complete frames - the P and B frames are not whole pictures - they only contain the difference between the previous frame and the next (P frame) or both (B frame) eventually you get to a point where you need another full image (I frame).

As I see it what you see in MEP is a reconstruction of the frames - ie if you take the sequence

I P P B I P P B I

What you see as you move from through the frames starting at the first   I   frame on the timeline is

Timeline frame:       1              2              3              4             5              6       . . . . . etc

 Image seen              I            I+P        I+P+P     I+P+P+B     I            I+P      . . . . . etc

Similarly when you play a movie on computer or TV you are actually seeing an illusion created from these three types of frames.

As an aside:-

This 'illusion' also exists in movies on film, these are complete discreet images (frames) and they exhibit a 'similar' problem.

If you look at the quality of them, when projected as still images on the big screen, they are very poor/grainy - it is only by projecting at 24fps or more that the picture quality appears to be very good - the graininess of the image is averaged out by the eye to appear to give a sharp image.

John

Hi John,

OK, I think I'm beginning to get some of it, albeit murkily! But I don't think this is relevant to the distinction I've raised. Because I've just done the same exercise on one of the original photos. No moving parts! That gave the same result.

So, PSP8 can eliminate a very uneven green background from that frame, while MEP cannot. With either Chromakey > Green or the Color tool. And the latter seems almost identical in function to PSP8's Magic Wand. Implication seems to be that the tools just aren't much good?

Practical bottom line question: do you think ironing and strong lighting is the only way I'm going to be able to achieve my aim?

--
Terry, East Grinstead, UK

Hi Terry

Back to the original question - the green screen - one of the things I did notice is that you are illuminated much more strongly than the green screen itself.

I can remember from my early days of digital photography that the green screen should be illuminated evenly and  lit so as to be about 1 stop overexposed compared to the subject.

Have you had a look at using a LED work light (see image) for illuminating the green screen?

Also good for the garage

John

Hi Terry and John

My 2 cents worth on this:-

I think you are closer to the mark Terry, in that regardless of what is going on in the image processing to create the frame we see, the effect is applied to a fully constructed image. John's analogy to film projection is not appropriate because that is about our eyes and brains creating an illusion by combining flashes and random grain patterns to create a homogenious picture through persistance of vision. Sure, this also happens when watching Video but for the purpose of your comparison the frame in the preview is fully formed as you point out, and all the process to get it there has gone on in the background and is a result of MPEG decoding, which is not an illusion but a process..

My understanding of how these editing programs work is that during preview the frames are being decoded happily from the compressed format (say MPEG2 or 4) in realtime, no small feat. When a transition comes along there are 2 frames that need to be decoded simultaneously doubling the processing load plus an additional load from any effect that might have been applied being carried out after decoding. It is conceivable that all this processing may mean that the effect's processing load may have to be simplified or slimmed down in order that the engine can cope and so as John has pointed out it may not be fair to compare a process within a program that is designed for a purely static single image. 

BTW John you have your B's and P's mixed up - it should be I B B P B B P B B I and yes it is a mind twister although remarkably, MEP2013 Manual has a surprisingly clear (laymans) explanation of GOP structure from page 317 in the MPEG Glossary. I find it a little confusing that P frames are called predictive frames because all they are doing is discarding information after analysis, that has not changed since the I frame...... but this probably refers to its actual order in the data stream rather than the GOP.

 Anyway we dont know in Terrys case whether his footage from his Webcam is MPEG?  DV or MJPEG for instance are intra frame codecs which means they have all the information in each frame albeit compressed, much the same as JPEG.  

"Practical bottom line question: do you think ironing and strong lighting is the only way I'm going to be able to achieve my aim?"

It will definitely help and the higher the quality and resolution of the footage the better. The pros use UV tubes (blacklight blue) to light the screen which glows a fluroescent lime green,because this is not a commonly ocurring colour in the subject and also prevents spill showing on the subject. 

Hi Peter,

Thanks, all makes sense to me.

As you imply, all that aspect of the discussion is now irrelevant anyway in view of my later test, on a still image. With hindsight I should have done that first, and not mentioned the movie clip! If MEP can't remove a colour from a still JPG then it obviously can't manage a video. 

BTW, you can just make out 'WMV' as the file type from the Logitech webcam in my first couple of screenshots, even with this site's stingy resolution.

It would be interesting to know what image editors others use, and whether (like PaintShop Pro) they too make light weather of removing my screen.

I've sent a formal support query about this, essentially asking if MEP's Chroma key functionality is up to scratch.

--------------------

I'd be grateful for thoughts on my post last night about Colour Correction please. I don't even know how you're supposed to use it properly - I've never seen a detailed step-by-step realistic example - but even from the depths of my ignorance it seems to get buggy after a few minutes intensive use. But the reason I was trying it was because I wonder if it has potential as a work around for my green screen problem? I've got as far as selecting it, so that it has vertical stripes, but what are you supposed to do next? Anything I do just removes the stripes...

--
Terry, East Grinstead, UK

Hi Peter

Thanks for the reference to the manual - I had not seen that one.

The pros use UV tubes (blacklight blue) to light the screen which glows a fluroescent lime green,because this is not a commonly ocurring colour in the subject

Agreed - in previous posts re green screen techniques in the old forum I lalways recommended getting a lime green colour cloth/paint for this purpose.

Most of us amateurs do not want to go to the high expense of using decent high power UV lights - not to mention the Health and Safety issue.

. . . not a commonly ocurring colour in the subject . . .  in humans depends if they are envious or not   - had another guy who was videoing try to break my video camera as he walked past me

John

Hi John,

Pretty expensive those lamps! If I get serious then maybe I'll have to make the investment.

But if I can knock up an image like this below in a few minutes from my cheap, unironed, unlit screen by using my old image editor, then IMO it points to MEP 2013's tools being below par.

--
Terry, East Grinstead, UK

Hi Terry

Pretty expensive those lamps!

The particular lamp in the image at £28 is a little more expensive than a halogen lamp, however they are bright, cool running and good for the energy bill / environment 

John

Hi Terry,

I would agree that MEP's chroma key adjustments leave a lot to be desired.  In addition to MEP, I am using Corel VideoStudio Pro X5 Ultimate and Cyberlink PowerDirector 11, and both products handle green screen much better than MEP.  Acutally the Cyberlink program puts the other two to absolute shame.  I was able to use the Cyberlink chroma key to fix the absolute worst case of what not to do with green screen from very bad lighting, to shadows to folds on the screen.  Neither of the other two programs could even come close to fixing that clip.  Does this mean that I like another program more than MEP? Absolutely not, neither of the other two have the logical workflow that suits me nor the applications stability that I need for my projects.

They both have some strengths over MEP so I use them to increas my capabilities.