the rules

Wow.Found here, a great trick, if a little uncontrollable. Start drawing a shape, say an arc or an ellipse. While the mouse button is down, keep dragging around and press ` (grave, under tilde, next to 1). Many shapes ensue.

If that's a bit random for you, create a shape, then apply Effect > Distort and Transform > Transform, rotate by 12°, 14 copies. Click the New button in the Graphic Styles panel to save that as a graphic style. You'll want a variation with 29 copies for asymmetrical objects or with 20 for something random in-between. Enjoy!

Adobe InDesign is a deep, rich application with a great many hidden and not-so-obvious features. It can even do some things with vectors that Illustrator can't. I've been teaching it (plus other Adobe and Apple apps) for a few years now, and I was asked to present at the recent meeting of the locaI InDesign User Group.

Here's the PDF from my InDesign Tips and Tricks presentation, with many pages of tips from all over the web, from other trainers and ones I found myself. If something's not clear from the pictures plus the notes, give it a Google or add a comment here.

Also here: the Table Transposition script referenced in the PDF and on the night. It exchanges rows and columns in the first table in a selected text box. To install, expand into the "Scripts Panel" folder in the "Scripts" folder next to the InDesign application, then find it in Window > Automation > Scripts. Works with CS2 and CS3 on Mac and PC.

Just a quick post on comparative speeds of various ports. The received (incorrect) wisdom in much of the PC world is that USB2 is faster than FireWire, because 480 Mbps is faster than 400 Mbps. However, potential maximum speed isn't the whole story. Because USB is a host-controlled bus, the CPU has to manage the data transfer while FireWire manages itself. Some good transfer rate/CPU requirements are found on this page.

Data for a Western Digital My Book Studio 500GB external drive with FW800, eSATA and USB connections; FW400 speeds are taken using a 9-pin (FW800) to 6-pin (FW400) cable:

USB2 averaged 29.0 MB/sec (15.0% CPU)
FW400 averaged 37.9 MB/sec (2.1% CPU)
FW800 averaged 61.5 MB/sec (3.7% CPU)

These numbers were (apparently) taken on a PC and historically USB2 has not been great on the Mac. Not sure if it has improved recently, but USB is always the poorest performer on a Mac. Probably more important is that you can daisy-chain one FireWire drive from another, so a single FireWire port can take as many drives as you care to throw at it. USB drives can't do this, so you'll need some powered hubs to connect many drives.

DV/HDV has lower data rate requirements than any of these ports should provide (~4MB/sec), but your minimum speed can never drop below the rate the video is arriving or you'll get dropped frames and likely an aborted capture. On the flipside is that some people report dropped frame problems with some Canon cameras and FW800 drives. I've had no issues capturing from my HV20 to my My Book Studio 1GB through FW800, but your mileage could vary.

Short version:

To use the HV20 in progressive mode, use the HDV1080p25 Easy Setup, but before you edit, change field dominance of all captured clips to "None".

Loooong version:

I have a Canon HV20 (PAL version). It can shoot progressive scan footage (1080p25) but it stores it in an interlaced video stream (1080i50). Note that there's no pulldown involved, and that the footage looks great without any processing.

There are several different flavours of progressive footage out there; Canon calls this one 25PF as opposed to 25F on some of their other cameras. The process is very different for NTSC frame rates going to 1080p24, which involves external programs and additional processing stages. PAL's 25p workflow is easy by comparison but issues remain in Final Cut Pro.

The issues stem from the fact that Final Cut Pro doesn't treat this footage as progressive material. Progressive-as-interlaced footage is OK for broadcast and delivery, but during post production, ugly interlacing artifacts can be introduced, for transitions and speed changes in particular. A horizontal push slide transition is probably the quickest way to see interlacing artifacts.

So, how can we treat this footage as true progressive material? Natively, it's only possible to capture this footage as 1080i50. Changing the Easy Setup to 1080p25 doesn't change the capture preset, only the device control preset and the Edit to tape/PTV output video settings. Footage is captured in 1080i50 with Upper field dominance.

It's tempting to simply create a progressive (field dominance: none) sequence and add the footage. However, Final Cut simply field-doubles the existing "progressive-as-interlaced" footage. FCP thinks the original footage is interlaced, and as the manual says: "Interlaced clips added to a progressive sequence are deinterlaced during playback." (III-688).

Here are some screenshots showing the field doubling:





Note that it doesn't matter what codec the sequence uses -- HDV 1080i50, HDV 1080p25 or ProRes 422 1440x1080p -- field doubling will occur and can be easily seen at 100% in the Canvas. This occurs with luminance and is not an artifact of the low chroma resolution inherent to HDV and DV footage, nor of RT Extreme.

Finally, a solution:

1. Select your HDV clip(s) in the browser.

2. Scroll along to the field dominance column.

3. Control-click in one of the selected items where it says "Upper" and choose "None" from the popup.

If you now put one of these clips into a new ProRes 422 1440x1080 25p sequence, all will be well. No field doubling and the resulting movie can be exported directly to an HDV 1080i50 sequence if required.

Problems:

1. This doesn't work when "field dominance: none" clips are inserted directly into an HDV 1080p25 progressive sequence -- field doubling still occurs. [EDIT: Nope, HDV 1080p25 works fine as a codec. My fault. Scratch this.]

2. If you nest the ProRes progressive sequence directly into an HDV 1080i50 sequence, the interlacing lines will return.

3. Affiliate clips already in sequences are not affected by the field order change on the master clip, so every clip already in a sequence needs to be adjusted. You need to play the movie, not just move to a different frame, to update the image. This can be a tedious process via Item Properties.

There seems to be another field doubling issue when resizing HDV footage to SD DV. This is (likely) due to field order problems. Solution: if you export the ProRes sequence, you can import the resulting QuickTime movie into a DV sequence with no field doubling issues. Inserting the intermediate ProRes progressive sequence directly into a DV sequence will produce field doubling unless the DV sequence is also set to progressive.

So, it's possible to find a good workflow, but not as easy as it should be. Ideally, it would be possible to remove the field doubling de-interlacing effect in the same way that you can remove the field order filter. A capture preset that automatically sets field dominance would be valuable too.

Any feedback welcome -- and please tell me if I've missed something, basic or otherwise. Note, if you're in the US with an NTSC HV20, you can shoot 24p in a 30i stream, but the process is quite different. Look at 1080p24 from HV20 (NTSC) workflow for starters.

You may that HDV is a bad format for editing; I've heard it from a few different places. Why? Long conform times before final output, effectively like you've put an effect over the whole movie and need to render it all. Also a reduced colour space, though no different to DV PAL. (Larry Jordan says as much, but look for Graeme Nattress's comment near the bottom.) You may also hear a recommendation to re-encode as DVCProHD or ProRes 422. Don't!

Import as HDV. Edit as HDV. Apply transitions with native HDV, but don't bother rendering unless you have to. You'll get the image quality of an online for editing with the storage space of DV.

If you're going to colour correct and actually finish the thing for broadcast, transcode now, at the end of the process, to ProRes 422 or some other lightly compressed format. You'll get better quality results (recompressing to HDV can indeed be ugly) but you won't have to capture the whole thing in ProRes 422 (at 4x the space) or DVCProHD (at reduced quality).

The golden rule: you can't get better quality video than you captured. Transcoding never increases image quality, though it can increase the stability of the existing image quality.

The exception to the rule: capturing live video from the HDMI port of many HDV cameras avoids HDV compression entirely, giving you a full colour space free from artifacts. The problem is that you're tethered to a computer with fast hard drives and/or an AJA I/O box. There is no point if you've already recorded to HDV tape.

The good news: Canon's HV 20 captures 1080p25 in a 1080i50 stream, with cine-like gamma for increased dynamic range, for AU$1999 RRP. Fantastic image quality for a (comparatively) tiny price.

Sample images from the HV 20, shot by me at home in Australia. No extra lighting or lenses.

http://funwithstuff.com/pics/kookaburra1.jpg
http://funwithstuff.com/pics/kookaburra2.jpg
http://funwithstuff.com/pics/possum1.jpg

I'll get to my web rules eventually, but when I do, I expect to be cribbing a chunk of Apple's Web Page Development: Best Practices. My major point of difference so far is that I'd recommend against designing pages that need tables to make them work. Enjoy!

use shortcuts

Most menu commands also have shortcuts, normally using modifier keys (shift, control, option, command). Almost every tool has a shortcut key, no modifiers required. Once you know the most used keys, you'll become much faster. V, B, M, D, X, Q, command-I, command-shift-I, command-A, and of course command-S. Another big one is to command-click a layer, channel or path thumbnail to load it as a selection.

apply all kinds of masks

A mask is an channel-based representation of a selection, usually controlling which parts of an image are visible. A selection is rough; the "marching ants" only show you the parts of an image that are 50% selected or greater, but a mask will show you more subtle information. Masks mean you'll never (well, 99% never) use the Eraser again. Quick Mask (simply press Q) converts a selection into a mask, and is the easiest way to convert between a selection and a channel. Once you've made a selection (perhaps using Quick Mask) you can convert it into a Layer Mask by pressing the dedicated button at the bottom of the Layers palette. Why is this so important? When you use the eraser, you're not only setting the opacity of a pixel to 0, you're effectively clearing whatever colour information was once there — because there's no "uneraser". When you erase by painting on a layer mask, you're only setting opacity to 0. A quick press of X to reverse black and white means you can paint the layer back in. Also very handy is the Clipping Mask, using the transparency information of one layer to define which parts of the layer(s) above are visible. Option-click between two layer thumbnails or choose Layer > Apply Clipping Mask.

definitely use smart objects (and smart filters)

As of CS3, Photoshop lets you do wonderful things with Smart Objects. Any number of layers can be converted into a smart object — essentially, a document within a document. A double-click on the smart object in the Layers palette opens it as an embedded document, much like an option-double-click on a placed InDesign document in InDesign. The joy of CS3, though, is that you can apply filters (and the Shadow/Highlight adjustment!) directly and non-destructively to Smart Objects. Especially powerful when you resize an image for web and find your sharpening adjustment is still appropriate.

use adjustment layers

Don't ever use Image > Adjustments unless you're editing masks, you're using Shadow/Highlight on a Smart Object, or you're using a strange adjustment that isn't available as an Adjustment Layer. Plain adjustments are destructive and of poorer quality than using Layer Adjustments. Why? Every time you apply an adjustment, you change the pixel values in your image. If you use Image > Adjustments, any fractional numbers in the result (very common) are rounded off. Adjustment Layers, though, are all applied in one pass, then the final result is rounded off — crucial if you use multiple adjustments, such as Levels + Curves + Hue/Saturation. Much higher quality, built-in masking, and far more flexible as you can edit and re-edit their settings indefinitely. Finally, one of the inconveniences of Adjustment Layers has been dealt with in CS3 — cloning and healing can be set to ignore adjustment layers entirely, avoiding "double-adjustments".

respect your image

I don't mean to get all floaty and nebulous, but remember not to push your image too far. If you're not careful, if you use destructive adjustments, if you don't use new layers for each change, if you don't avoid clipping, if you push your image too far in Camera RAW... your image will suffer. Take the best image you can, and use the image itself to provide your selections, through channels. When you're retouching, put all your adjustments on separate, new layers, so you can selectively remove parts of your image. And please, please, try not to make everyone totally "perfect". We're not, and it makes your job harder. Retouching, like any lie, is easier to pull off if it's close to the truth.

use a graphics tablet

Because you can't paint a mask properly without one. Clipping paths are easier too. A tablet can be good for avoiding RSI, as it adds variety to your arm movements. A tablet also lets you make fuller use of Photoshop's brushes, but note that if you're painting straight lines (common when touching up masks) you'll actually want a non-pressure sensitive brush to avoid creating comets. Fancier tablets allow you to change brush characteristics based on brush pressure, angle, rotation and more. Cheaper tablets only recognise pressure, though they do come with a built-in pen holder.

know your brushes

First off, brush heads are very useful (and you can make your own) but the basic brushes are most flexible. With a normal circular brush selected, use the square brackets [ and ] to change the brush size, and shift-square brackets to change its hardness. When that's not enough, remember that there's a big difference between changing the brush preset and simply changing the brush head — you'll miss all the pre-built changes in scattering, colour, shape and other dynamics. Scattering is often useful, as is random brush head rotation on natural media brushes. Also remember that brushes aren't just used by the brush tool; the Art History and Smudge tools can use good brush presets to great effect. Finally, if your brushes are stuttering, lower the Spacing value.

dodge and burn to clean masks

When touching up masks, it's natural to use the brush tool to paint in black and white. Be careful the brush isn't too hard or soft (shift [ and ] ). That said, if you need to clean up the edge of a mask, the brush is the wrong tool. Set the dodge tool to work on highlights and the burn tool to work on shadows. Now you can paint as close as you like to the edges of a mask with ease.

use fullscreen mode

Often you need to work on multiple files at once, to drag between them, to compare one file to another. But if you don't and you're clicking anywhere near the edge of a file, full screen mode gives you the ability to click near the edge of the canvas with a large, fuzzy brush. Focusing only on the image (f, f, tab) also gives you the ability to ignore the interface, to ignore your desktop background, to become one with the image. Om.

make friends with RGB

Yes, if you're printing, you're printing in CMYK. But very likely, the image originated in RGB. CMYK is an output format, and if you do all your colour correction in CMYK, you're missing out on the greater colour gamut (and thereby quality) of RGB. Every output device is a little different, and leaving an image in RGB as long as possible will maintain higher quality on a wider range of devices. Make sure your colour profiles are accurate, and use RGB for as long as you can. If you need to perform final adjustments in CMYK, go for it, but keep your fully-layered RGB file too.