[Steve Bentley]

What you want is to get is the text’s length. If you are using CC2014 or above you can use
myTextShape = thisComp.layer("text").sourceRectAtTime(time,false);
myWidth=myTextShape.width;
myHeight=myTextShape.height You probably don’t need the last line, but you never know.

Then you can position the end of the red bar at text start pos+myWidth-(barlength/2)
This assumes the bar’s anchor point is centered.

If you are using a version of AE before 2014 then it’s much harder. And thats due to the fact that the spacing between characters differs between all fonts and all characters within a font. (except monospaced fonts). The old version of AE doesn’t have the SourceRectatTime class to access.

In pre2014 you can get close by using

myLength= thisComp.layer("text").text.sourceText;

myWidth = myLength.length;

But you wil have to put a multiplier or a slider effect on the next line to account for the differences betweenn characters and even then it won’t be perfect.
adjustedWidth = 100 * (myWidth * .175)+ thisComp.layer("sliderLayer").effect("Slider Control")("Slider");

[Steve Bentley]

Also watch out for CMYK vs RGB when you are in AI. Make sure to convert all files to be used in AE to RGB.

[Steve Bentley]

Can you use expressions? You could poll the length of the name layer and then apply that value to the position of the red bar, assuming that red bar can have one end hanging off the end of the frame width and be as long as the frame is wide.

[Steve Bentley]

I’m not sure what you are seeing but you can just apply the fill effect to fill a mask with color. You can feather the edges of the mask, invert it, mask it with other masks, all within the mask controls.

[Steve Bentley]

In your compression software you can set how far each keyframe is apart – I think AME defaults to 5 seconds between keyframes (so 150 frames at 30fps) To my thinking this would only be appropriate for watching paint dry or grass grow because you just couldn’t have much change (action) and have smooth clean video with the key frames that far apart. But that’s my level of quality need, yours may vary.

So the trade off is – the closer the keyframes are the smoother the video is going to be (as in: dissolves or fast action wont shred or go pixely or blocky) but it also comes with a higher mb/s bandwidth. Moving the key frames farther apart uses less bandwidth (because the keyframes are where all the data is stored for the rest of the frames and farther apart means there are fewer of them- that’s overly simplified, but good enough for now). Because you have less bandwidth being used by more distant keyframes, and because your overall target bandwidth hasn’t changed, more of that available bandwidth can go towards the accuracy (and abundance) of colors.

You might have noticed that colors look a little less saturated after a compress. This is that balance being maintained by your compression software. It needed to steal some of the color bandwidth to make sure the motion was smooth. (again, overly simplified)

By the same token if you have a file that is playing at 30fps it will naturally need more data than a file that is playing at 24fps, so again if you can render a slower frame rate, the bandwidth that would have normally gone into creating many more frames per second, can instead be used for more accurate colors. For the web (usually) it doesn’t really matter what the frame rate is. Anything above 18 is going to look like real motion (half the time the web is throttling the frame rate anyway to something lower or intermittent).

But you can’t just say “please play this 30 fps movie at 18fps” because it will look like it’s in slow motion) you have time-interpolate to a new frame rate so that an action that took 1 second at 30fps will still take 1 sec at your new frame rate.
Even better is to plan on a slower frame rate in your creation. (like 24 or 25)
If all elements in after effects have been moved or transformed in AE and there is no running source footage, you can simply do a time stretch on a precomped version of your main comp to get a different frame rate. AE will interpolate that perfectly. If on the other hand you use running footage with discreet frames, you must use something more complicated like the built in time warp effect or RealViz twixtor to do a proper change in the frame rate.

You can often tell your compression software to do the time warp for you, but unless it’s a perfect division like 30fps down to 15 or 10 fps, it never looks very good.

[Steve Bentley]

Right, the cell shader’s thickness options leave a lot to be desired (like… everything to be desired, but it does do it cleanly).
So what we’ve done in the past is to do a depth pass or position pass and use that to fade the lines as they recede adding dimension to an otherwise very boring line.

The ST shader is known for “boiling” especially when you animate the lines so I assumed you were animating them in the shader.

Its all about angle. ST knows to put a line when the angle of view or angle of the surface normals exceed or fall below a certain point which you can set. But its tricky and with an animated object might always be changing. So you could try eliminating some of the line types in either the texture tag or the over all render settings. Cusps and folds are the trickiest to get consistent so turn off all line types and test render each one and see which ones you can do without. Or if you have time you could render each line type needed as a separate pass (you can do this now in multipass) and then dissolve in each one within your compositor as you need them and dissolve them away just before there is a hiccup.

You can often change your start frame and render multiple copies, which will give different results for each render, then overlay them and blend them together to get a more consistent whole.

[Steve Bentley]

First, watch the size of the screen – if it gets too small it will moire with the pixels of the video itself. The dots can also often sparkle if they get too small and the sub sampling changes slightly from frame to frame. If you ignore the sparkling stars in that first video you can see this happening with their screen over the space areas.
A dot pattern that is slightly fuzzy will work better than a super sharp clean vector version. And an oversized dot pattern that you scale down in AE will help with the subpixel sampling and reduce sparkle and moire.

As for the effect itself, try changing the blending method. Instead of using the dots as an alpha try overlaying the dots using multiply or hardlight or exclude. This will allow the dots to react to the level of light and color that is in the scene already, naturally giving areas where there are dots and areas where there arent. There is a keyboard shortcut that lets you quickly change that blend effect and see what you like.

You could also premask the dots with a fractal noise so that some areas show and some don’t.,
You can also play with exposure on an overall adjustment layer so that some areas “bloom” and the dots wash out in areas of high brightness. Combine these two techniques to mask the adjustment layer with the fractal noise or soft masks to control what areas bloom.

Don’t make the dots 100 percent opaque.

[Steve Bentley]

Dave’s correct – render out of AE as close to uncompressed as you can and then compress to H264 in something else (like AME). You can’t really do multipass compress in AE, and AME (which can) will give you more bang for you buck (color per MB wise). AME will look at all the frames first and say “hey, there’s a lot of this glow thing going on, I better give more priority to that”.

Are your glows blue? There are less blues in the color gamut your monitor uses compared to any of the other colors, so banding shows up the worst for blue because there are so few blues to choose from; the color has to “step” to the next one as it graduates. Add a cross dissolve to that blue glow and now you are really asking for it. Big changes in image need the keyframes closer together to capture those changes cleanly, and that adds up to more mb/s.

Unless you are looking at your movie on a color grading monitor that can do 10 or 12bit color, it’s always being converted to 8 bit color even if the originating file is 16bit. But talking about H264 and 16bit color in the same conversation is a bit like comparing an aircraft carrier to the toy boat floating in your bathtub. One is trying to hold an immense amount of data and one is trying to hold the least.

To get the kind of throughput you need for real time video across the web, H264 eliminates a bunch of data and that can lead to banding. The only ways around it are adding noise, or compressing to a higher Mb throughput (or a slower frame rate which is essentially the same thing – less fps, means less mb/s so more bandwidth is put toward the colors and motion than higher frames per second.)
Also, adding noise can actually increase your bandwith because now H264 is having to deal with all these little multicolored flickering pixels, instead of just the original problem.

Try it. Increase your encoding output Mb/s and see what you get. For a top end bench mark – bluerays are encoded at 25,000 -29,000 kb/s and even they have banding sometimes. You probably won’t see much of s difference between 15,000 kb/s and above and blue ray levels. Also try increasing the distance between keyframes as this will focus more of the bandwidth on the color rather than the action.

[Steve Bentley]

Right and we are talking about H264 here. I can export a QT .mov in H264 from AME, with multipass and (as far as I’ve tried )any resolution I like, included a non-16pixel imperfect-square division. (and we do some weird formats here, eg 23,040 x 1080, so we’ve done our best to test those resolution limits)

So given that I didn’t really understand your first sentence. I can export a QT from AME in a non apple codec with all those features. Unless you mean Apple has hijacked H264 and created their own flavour.

[Steve Bentley]

A couple of things to try apart from general color grading:

First, depending on the screen your audience will be viewing this on, the black levels can change dramatically. And I find this is where you first get the creeping feeling that the shot is fake. So temporarily jack up your black levels on your monitor and you may see big differences in the black levels of the robot vs the surrounding plate. (you can also do this with the info box and dragging your pointer around checking shadow levels) but changing your monitors gamma will make it obvious. If you match these levels you won’t see a difference in the shot on your monitor set to normal but it will survive the hideously large array of gammas out there now that we all have LCDs.

Second sneak trick is to add a lens flare to the whole shot – even if you don’t need one. You can position it off screen and let it fog in to the viewing area. You can either position it so that the lens artifacts are not visible or you can turn them off if you have a pro version of a lens flare package (I don’t think you can turn them off in the AE built in lens flare).
This has the same result as above – it fills in the differences in levels and makes them come out all the same.
So it shouldn’t look like a lens flare or that there is a bright light somewhere, you are just adding it for the fogging effect. It also reduces the perfection that CGI renders always have.