Too Much Too Soon HELP!

if I had a mac, I'd consider your workflow but this avisynth+vdub+AE+Premiere is just too many apps to work with. I'm learning how to uprez in instant HD, but hadn't had any time last night so I'm gonna try to do something in the next couple of days. It seems that when I go from 720p to 1080, it crops the image? Dunno, gotta work on it more.
 
so premiere can't open avisynth files? i thought they were universally recognizable as regular avi's. i haven't used premiere since 2000 though, whatever version that was.

/matt
 
Providing I do the Adobe Premiere road and then through Avisynth+Virtualdub for stairstep fix, what would be the best output setting to put into Avisynth/Virtual dub and then out again?

Obviously the less compressed and lossless as possible, so which is the better and free codec?

Does anyone know what would then be the best method, on a PC, to convert 24p to 25 pal broadcast standard and in a format to convert to DVD?
 
mattsand said:
so if you buy a razor, store it in a locker and never take it out, are you then blaming it that your beard grows? sorry honey, my razor hasn't shaved me. ;-) or to make the analogy better, if you take it out every day, wave it around in front of the mirror without letting it touch your face then put it back? you gotta do the work. tools, computer code even more so, have very little intelligence.

/matt
Click to expand...

Language Barrier kills your attempt at a good analogy. I'm just trying to figure out what I'm doing wrong, I didn't say that the plugin didn't work.

Quick to take offense.

So, again, I tried dropping a slug into the timeline, dropping the upscaler onto that slug, dropping the desired clip into the image well and rendering: nothing happened or at least I didn't notice any drastic changes.

Again, what is it that I might be doing wrong?

Thanks.
 
hmm, good question. i'll try that. i thought that you had to use avisynth file in vdub. matt, you suggested removing the last line, but all the #### or the squares, do they matter?
 
what language barrier? i mean english is not my native language and i obviously make mistakes, but i've lived in new york before, i went to a college where all literature was in english as well as most of the professors, and i've been on the web since forever and never had much trouble understanding or making myself understood. i've read my post over and over again and can't for my life understand what's hard to understand, which must mean that you think i didn't understand you? i *know* you didn't say the plugin didn't work and i *didn't* take offense. you said that the plugin didn't do anything, hence my analogy, it won't unless you make it. i'm just trying to explain that you gotta learn how to use something and i think you're letting go way too easily. i mean i'm not the one who suffers when you don't get the results you need, am i?

so did you drop the clip onto the "source clip" well in the filter control or into the timeline?

/matt
 
Car3o said:
hmm, good question. i'll try that. i thought that you had to use avisynth file in vdub
Click to expand...
they are frame server files with an avi interface. when the app requests a frame of video, the script is run and the result fed to the app just as if it was a frame from a regular avi file. they should work in any video for windows app, though not necessarily in all directshow ones i guess.
matt, you suggested removing the last line
Click to expand...
the last line *of code* not necessarily the last line of the text file. the part that tells the script to scale the image, nothing else. you can also comment it out using a # before it.

/matt
 
mattsand said:
what language barrier? i mean english is not my native language and i obviously make mistakes, but i've lived in new york before, i went to a college where all literature was in english as well as most of the professors, and i've been on the web since forever and never had much trouble understanding or making myself understood. i've read my post over and over again and can't for my life understand what's hard to understand, which must mean that you think i didn't understand you? i *know* you didn't say the plugin didn't work and i *didn't* take offense. you said that the plugin didn't do anything, hence my analogy, it won't unless you make it. i'm just trying to explain that you gotta learn how to use something and i think you're letting go way too easily. i mean i'm not the one who suffers when you don't get the results you need, am i?

so did you drop the clip onto the "source clip" well in the filter control or into the timeline?

/matt
Click to expand...

That's my misunderstanding, then. I thought you were offended and snapping back at me! I just couldn't figure out what was going wrong or what I was missing.

I dropped the clip into Source Clip and checked it on the timeline, rendered an unrendered. Nothing. Flipping back and forth through the monitors, etc, showed no real change in the footage.

However, since my Tower is a dual boot machine (booting natively into XP or OS X) I flipped over to Windows and tried the VirtualDub process and it worked right out of the gate, albeit I exported to uncompressed which showed a large shift in contrast... I assume it's because I'm lookin' at it in Quicktime.

I can definitely see cleaner edges and it looks like the banding actually got better. That statement pending a visual combover.

Does the FCP workflow require a certain project setup or anything else that I may be missing?

Thanks for taking your time to answer all of these questions and in writing the scripts. I saw that it worked for others, I just can't figure out what I'm doing wrong personally.
 
so are there any jaggies in the source clip and are they still there when run through the upscaler? it sounds very unlikely since to get an image at all the plugin needs to run, otherwise you'd just get a black frame, the slug. the best way to compare is to place the clip on v1, the slug on v2, drag the clip to the well, and toggle v2 on and off, making sure you're at 100%. does the rescaler work? you mention scaling to fit, there's no such thing involved, the upscaler does that automatically. have you already scaled the source clip before dropping it in the well? that breaks the cadence of the stairsteps and thus the algorithm.

as for the banding if you're using the picvideo codec with advanced deblocking you do get a rather big improvement compared to all other codecs, apple's included. i transcode all my avi's into 100% mjpegs using virtualdub before importing them to fcp (actually i offline first then use media manager to copy only the used ones), you can also use huffyuv for a completely lossless conversion.

/matt
 
You're definitely right about that. When I went back to OSX I noticed that the banding returned. Funny! But, even funnier is that it was working all along and I think I just expected a large increase in image quality... fact is that I forgot that beckspace's example also included neat video for denoising.

Your method actually works better because it doesn't futz with the contrast and you can use it natively in FCP.

So, in short, this is pretty easy and it works, now I just have to get those guys over at NeatVideo to re-assign my serial to CS3/OSX.


Thanks for the tips on transcoding/converting, as well. Doing straight AVI didn't seem very plausible to me. Looks about 10 percent better this way, I believe neatvideo will help and instantHD if you wanted to go larger.

I still can't get over BeckSpace's video, though. That stuff was crazy clean!
 
neatvideo is great for saving footage that's not supposed to be grainy, for matching, or keying, but it sucks as a general method. the result looks way too clean, it removes a lot of real texture no matter what they're trying to tell you and the d90 has already removed a lot with it's own noise reduction, crappy debayer algorithm and heavy compression, and workflow-wise it's a pain with the rendering time. just shoot clean footage to begin with. ;-)

you can definitely try instant hd, i use it myself, but it doesn't actually make a big difference. it's only a 30% scale from 801 to 1080, remember.

the avisynth script and my plugins do *exactly* the same thing, the results all else being equal are bit for bit identical. if you see a difference it's always from some other source.

/matt
 
I think you're right about seeing a difference from a different source. The jaggies are gone just the same when I lay one clip over the other, looks no different.

So, yeah, shoot clean footage to begin with or get a 5D for super lowlight production.
 
Kholi said:
You're definitely right about that. When I went back to OSX I noticed that the banding returned. Funny! But, even funnier is that it was working all along and I think I just expected a large increase in image quality... fact is that I forgot that beckspace's example also included neat video for denoising.

Your method actually works better because it doesn't futz with the contrast and you can use it natively in FCP.

So, in short, this is pretty easy and it works, now I just have to get those guys over at NeatVideo to re-assign my serial to CS3/OSX.


Thanks for the tips on transcoding/converting, as well. Doing straight AVI didn't seem very plausible to me. Looks about 10 percent better this way, I believe neatvideo will help and instantHD if you wanted to go larger.

I still can't get over BeckSpace's video, though. That stuff was crazy clean!
Click to expand...


which beckspace video are you referring to?
 
Original
Original.jpg


Avisynth
Avisynth.jpg
 
It is the codec. Avisynth uses video for windows so it may also use another codec depending on how you set it up. It can also be the host app but definitely not the script. 1280x801 is an intermediate format that contains the most amount of information with the least amount of manipulation which you then scale to your preferred size.
 
You must log in or register to reply here.
Back
Top