[Dave Haynie]

So.. this got me curious. Just what IS the impact of image size. Having 100’s of 18Mpixel shots around from a current photo project, I tried it. I created one 5min project with the 18Mpixel images, 5sec per image, and the same one with images resized by Lightroom to be 1080 pixels high. No crossfades, nothing but JPEG (since Lightroom doesn’t batch-convert PNGs) just image after image for 5 minutes. I render using Sony’s Blu-ray 1080/24p 16Mb/s preset. GPU rendering is on… I didn’t see enough GPU activity to worry about A/B-ing it.

Good/8-bit Best/8-bit Best/32-bit
Large 7:22 7:21 10:41
Small 7:10 7:06 10:37

So… yawn… not much of a win here in pre-resizing. There’s a difference, but it’s very insignificant compared to all the other work involved. There’s nothing shouting out quality-wise, either, though I didn’t do a photo-nerd-style examination of close up features. Maybe some other time.

So I did a slow zoom, so Vegas has to re-calculate every frame. In this, the large project has an 18Mpixel PNG, in the small project, that same image in PNG, resized to 4.5Mpixel. This is a zoom out for 1 minute. On the small project, the video starts at a zoomed-in height of 1080 pixels; on the large project, the start position is scaled deliver the same image size, but obviously with four times as many pixels.

Good/8-bit Best/8-bit Best/32-bit
Large 1:41 1:41 3:08
Large noGPU 1:54 1:59 3:12
Small 1:31 1:32 2:52
Small noGPU 1:43 1:49 3:04

Here I saw the GPU busy, so I decided to run both cases. This probably explains why “Good” and “Best” are tied — it’s all overhead, when you use the GPU for scaling. The actual math is fast enough to not be independently measurable. You could claim that the GPU case uses the same scaling algorithm in both cases, and while that’s possible, I have seen other cases in which “Good” vs. “Best” make a difference with the GPU.

Bottom line: if you’re concerned about highest quality, maybe you want to resize your images. If you’re on a low resource computer, also a good idea. But I think you have to go beyond Photoshop for any significant difference in quality, and for performance, you’re going to need a long and complex project before photo resizing makes a big difference.

I should point out a related issue I didn’t benchmark here: drive speed. If you have dozens of 18Mpixel images in the same project, you may kill your HDD performance and bottleneck there, not in the actual rendering pipeline. But this is more related to file size; 18Mpixel JPGs might well render faster than 2Mpixel PNGs, when you have very large numbers. And enough RAM.

I should point out that there has been some evolution here for sure. There was a time when Vegas had issues with larger photo files… and that was back in the 6-8Mpixel days of DSLRs. And Vegas 7.

-Dave

[Dave Haynie]

[Bob Peterson] “Let me expand. In Photoshop, the recommended method for decreasing a photo’s size is bicubic sharper while the recommended method for increasing a photo’s size is bicubic smoother. Unless, I’ve missed something, Vegas lacks this degree of sophistication. It can certainly resize, but it does not produce the same level of quality.”

Ok, Vegas is far from ideal for upscaling photos. Photoshop is no great shakes, either. The best blowup you’re going to get is with specialized tools, though I’m fairly convinced that stair-interpolating in Photoshop does enlargements better than simply one-shot resizing. Surprisingly, Photoshop (as of CS4 anyway) doesn’t offer any of the more modern algorithms. You’ll get even better results on most material with windowed sinc (Lanczos 3) or spline resizing.. you can get sinc in The Gimp, InfranView, Virtualdub, AVISynth, etc. And there’s some argument for the various proprietary upscalers, such as reshade, BlowUp, or Perfect Resize (nee Genuine Fractals). But this is largely a print photograpy problem. If I have a small bitmap I need larger for video, I’m probably going to be unhappy with any upscaling alone. So I’m going to add blurring, art effects, something in my bag of tricks to hide the fact it’s a lousy image.

Vegas is only really designed to do down-scaling well. Which makes sense; you don’t usually upscale video, and when you do (eg, SD material into HD), you really want specialized processing. Vegas controls this with the “Full resolution rendering quality” setting in your project preferences. If you set “Good”, you get bilinear scaling, which is fast. If you set “Best”, you get bicubic scaling (“with integration”), not so fast, but better. And pretty similar to what Photoshop is going to give you. Sinc, spline, gaussian, etc. probably produce a better looking result, at the cost of time.

This is more applicable to Vegas anyway; Photoshop is NOT optimized for small image manipulation, it’s optimized for print work, lots of pixels. I have 750MB .PSD files around here. Vegas is pretty much only concerned with 8Mpixel and below (and usually, 2Mpixel and below), but also speed. The other thing about video; you don’t really have the time to spot small differences in resizing when displayed at 24-60fps. I assume that’s why Vegas has never bothered with a kernel more sophisticated than bicubic — you might see the difference, zoomed down in a photo editor. But on-screen? I’m personally not sure.

Of course, if you have the time, there’s the Content Aware resizing tool in Photoshop, which can be kind of magical… particular in dealing with aspect-ratio problems (there’s a plug-in for The Gimp that does similar things). It’s ok if you’re resizing a couple of photos, and it will produce a superior result. But for a large number in a complex video project, I’m probably not going there.

[Bob Peterson] “Certainly, if panning or zooming is needed, Vegas should have a larger image to accommodate that function, but routinely giving it 18 megapixel photos strikes me as a recipe of extremely long render times.
“

Certainly, for lots of static images, there’s no point in not resizing elsewhere just to make Vegas run faster. It depends quite a bit on how much you’re zooming/panning, and how much you’re using stills. If it’s all stills, I’ll bet the scaling time becomes significant enough (though still less work than, say, decoding AVC); if just a little here and there, I’m wasting time if I need a special step to resize in Photoshop.

Truth of the matter is, anything from my camera was shot in 18Mpixel RAW, and even though Vegas 11 supports RAW, that’s kind of joke, since they use Microsoft’s RAW decoder, which is like a decade out of date. So I’m almost certainly converting from RAW and/or 16-bit TIFF, selecting photos in Lightroom, which uses pretty much the same adaptive bicubic resizing algorithm that’s built-in on Photoshop.

-Dave

[Dave Haynie]

Yeah, the full version of Nero will definitely let you make and burn a DVD from TM700 video. My only concern is the HD handling. The story used to be that you bought the full version, then paid for a plug-in for HD support. But I believe that HD extra is only needed to create HD materials or play-back actual Blu-ray discs. I that’s what they call “Platinum” now. So the normal “Nero 11” ought to do what you’re looking for, creating an SD-DVD from HD sources.

They do have a time-limited demo for download. I’d definitely check that out before paying.

-Dave

[Dave Haynie]

As a video editor, Vegas can absolutely be expected to resize a photo just as well as Photoshop. The problem is that Photoshop has, effectively, all day to do it, and when it’s resizing, that’s all it has to worry about. Vegas is resizing photos, maybe video, compositing, etc. And doing so where speed really matters. If you’re doing this in a system with more limited resources, they may get spread pretty thin with all this going on.

Pre-scaling your photos is one way to offload this work — but don’t go overboard. If you’re panning and zooming within a photo, keep it fairly high resolution. And at least the recent versions, Vegas 9/10/11 on 64-bit systems with plenty of memory, handle my 18MPixel JPEGs and PNGs just dandy, no absolute need to resize anything.

-Dave

[Dave Haynie]

[Dave Petteruto] “s there some sort of inexpensive software that will achieve this task, Nero maybe?”

Nero Vision will do this.. it’s part of the Nero 10/11 suite. You don’t get this if you just buy Nero Burning ROM. The basic Nero 11 suite costs $79.99… I’m not certain that it supports HD intput, it definitely only does SD output, but that’s all you care about. Vision is a very, very bare-bones editor. You can drop in a file (=clip), make a few cuts, and yes, insert chapter marks. Then render it to DVD. I actually have a full version of Nero 10, including HD support, so I can’t test against plain Nero. But they do have a free trial. I dropped some 1080/60i files from the TM700 in, Nero handled them just dandy.

Of course, Vegas Movie Studio HD Platinum 11 is only about $15 more, and will offer dramatically superior editing. Not what you’re looking for today, but worth considering.

There are cheaper options, such as Corel’s DVD Movie Factory, which is $49.95 this month. I have not used this, but it is cheap, easy, and they also have a free trial.

-Dave

[Dave Haynie]

I might well use the masking technique suggested… just mask off the portions you’re not using. Another technique is to simply draw the flight paths. Well, of course, you’ve already drawn them… so what you want to do is un-draw them. This is a pretty easy animation technique. You start with the drawn route, then save off successive PNGs, each with a little more erased in Photoshop.

In case that’s not clear, you can see this kind of thing in an animation I did earlier this year (below). Each time something is being “drawn”, it’s probably, in reality, being erased in reverse. Masking is easier if you have one straight line, since Vegas will do the “tweening” for you — you only need to keyframe the start of the mask and the end. But if it’s too complex, just animating the line might be the easier thing.

https://www.youtube.com/watch?v=lCrUrOQkJSw

-Dave

[Dave Haynie]

[Daniel Cervantes] “Is Vegas 7.0 not Jpeg friendly? I save the file by moving the file the Jpegs were in and leaving them out of the vid. Thoughts? Maybe resave the Jpegs to a lesser definition?
“

Probably a good idea if they’re from a modern camera.

Vegas has progressively improved on its still image handling. I use these features extensively: for animation, for “Ken Burns” style video, etc.

Back in the day, around the Vegas 7 era, I found that a large number of direct-from-DSLR JPEGs could bring Vegas to its knees, and that, when on its knees, it was much more likely to crash. First thing I did was put the JPEGs on a drive other than that used for video (and in fact, video input and video output went to different drives). This dramatically improved rendering speed and reliability, at least in my case.

Second thing… yeah, sometimes, I could shrink the images without worries. In the event of a fixed image that’s not panned or zoomed (rare), you can’t use more than 1920×1080 pixels, anyway. So you can resize any photos that aren’t moved around that way. And then, given today’s much larger camera sensors, the problems are just that much more likely — the DSLRs back in 2006 or so were 6-8Mpixel.

-Dave

[Dave Haynie]

I wasn’t recommending using Vegas just yet. If you want to merge the files without recompressing, Vegas ain’t the answer.

For simple concatenation, I just concatenate the files. Specifically, I open a cygwin shell, and type something like this:

> cat /cygdrive/j/00000.mts /cygdrive/j/00001.mts > /cygdrive/d/merged.mts

I think you can get something like that in a regular Windows shell doing this:

> copy J:/00000.mts+J:/00001.mts D:/merged.mts

Again, this will always work if you’re merging transport stream files, OR files broken at exact 4GB or 2GB boundaries. Another option is a little Windows shell extension from JD Design called “File Concatenate/Split Context Menu”. This lets you do the same thing in the Windows Explorer — select a few files, pull down the “Concatenate” menu, and this pops up a dialog that lets you order inputs and select an output file. See here: https://www.jddesign.co.uk/products/concat/concat-s.htm

The other option is to use a smarter file/merge utility. This is recommended for joining files that are logically split by the camcorder (you’ll notice that every file is a slightly different size if the camera is doing this, rather than all the files being 2GiB or 4GiB). I suggested YAMB (see https://yamb.unite-video.com/download.html), since it’s designed to processo MPEG-4 files, and it’s what I use for merging arbitary MP4 files.

-Dave

[Dave Haynie]

[nelson araujo] “and if someone could explain why the video plays smoothly on WMP and not on vegas i would very much like to know it , even if it doesn´t help at least i´ll be glad to understand it.”

Basically, you have a pretty kick-ass GPU (I have the same one) and a relative weak (by this week’s four-core standards) CPU. When you play a video in WMP, it may be able to tap into video-specific GPU acceleration. Particularly, if you’re running Windows 7, you have an AVC decoder from Microsoft that uses the DXVA 2.0 interface to do much of the decoding on the GPU. So on my system, for example, I can use something like 60% of my hex-core AMD1090T to decode a 1080/60p AVC stream in realtime. Or, using GPU acceleration, easily do four or five of these at the same time.

Vegas 11 add some GPU acceleration (if you enable it, of course), and with the latest AMD drivers, you’ll get to use that. Make sure you’re using this… download something like GPU Shark to check your GPU use, then make it’s being used during playback.

It doesn’t have a profound effect on a single video stream, however — they’re still mostly using the CPU for decode, the GPU more for compositing and effects. I have seem some instances were preview goes 3x+ faster with the GPU acceleration in play.

You can make some general improvements in editing speed. In your project preferences, make sure you’re editing in 8-bit pixels, and have the quality set to “Good” or lower. And of course, you can adjust the preview windows itself in terms of quality and size of the preview, to speed things up.

-Dave

[Dave Haynie]

I did a clone shot a couple of years ago, for a music video (local radio station contest). Just one clone, and I didn’t have the time (or stones) to attempt those crossovers… very cool. My main issue was shooting it outdoors — I had to worry about wind and changes in light. So it was tricky editing.

Also, very time constrained… I hadn’t actually finished the song when I had to shoot the video. Made it with a few minutes to spare!

See here (clone comes in around 0:57):
https://www.youtube.com/watch?v=UNmfm-BAwzs

-Dave

Some contents or functionalities here are not available due to your cookie preferences!

This happens because the functionality/content marked as “Google Youtube” uses cookies that you choosed to keep disabled. In order to view this content or use this functionality, please enable cookies: click here to open your cookie preferences.