RU.08

Please do not use the ending that the director hated and fought against:

“I didn’t want to have the alien come out. I still don’t like the idea of the alien emerging. Originally she falls backwards, standing on the gantry, with an explosion of blood on her chest and this thing pushes out. It’s more a stigmata and she falls backwards into it.

“I never felt it was necessary to show the creature. I was very much against this and dragged my feet and said ‘I don’t believe in it, I don’t think it’s important to see the monster but if we’re going to do this we’re going to have to do something that has a little bit of top spin to it, something else going for it.

“No matter what cathartic experience we could expect from finally seeing the two strongest images from the first movie, the chest-burster and the character of Ripley, if we left the movie with her choking on her tongue they would feel worse going out of the film than they do now.

“I said ‘whatever happens she has to be in peace at the end.’ It has to be a sigh rather than gritting teeth and sweat. So we talked about it and went over and shot this blue-screen element. We were shooting that shot four days before the film opened, a completely ridiculous mess. I don’t know if it works.”

Asteroid-Man said:

Ignore storage, I'm talking about uploading 10gb at once. Once it is stored, people downloading it only uses the download bandwidth of those accessing the files. If Mega considers that the same as upload bandwidth they're barking. I worked in digital distribution on big shows, my friend. This is something I know quite well. ;)

50GB storage, uploads don't count against the bandwidth only downloads from the cloud storage count towards their "10GB per month". As other have said they seem to allow more than the stated 10GB bandwidth per month, but I'll bet you once you get towards your 50GB storage limit that they'll start to throttle the bandwidth more to push you onto a paid plan.

Also, if you share the links publicly they will take down material when a DMCA complaint is sent to them, so you should only share the links through PMs.

-V

It's on page 19 of the document I linked to Frink, which is on the Mega site. I believe you're right towne32 that the bandwidth does usually reset sooner, but their stated policy is 10GB/mth on the free account.

Mega is limited to 10GB of bandwidth per month though on the free account.

I suggest just fixing the megashares links for the DLC package - or putting in myspleen. If you fix the DLC then I or someone else can up it to the spleen for you.

And... just because I can, lol:

-V

towne32 said:

Valeyard: do you know if there is an avisynth script available that does the equivalent of vidfire? Would be curious to try it on the underwater menace 2 episode now that the DVD is cancelled. Have a fairly decent quality copy of it.

Here you go, I just tested this script and it works. Please note this assumes you don't want to resize the video (e.g. going from 576p to 576i),l I used SVP for the frame-doubling:

even=last.converttoyuy2().separatefields().selecteven()
threads=10
super_params="{pel:2,gpu:0}"
analyse_params="""{block:{w:8,h:8,overlap:3},
    main:{search:{coarse:{distance:-10}}},
    refine:[{thsad:200}]}"""
smoothfps_params="{algo:21}"
super = SVSuper(super_params)
vectors = SVAnalyse(super, analyse_params)
SVSmoothFps(super, vectors, smoothfps_params,url="www.svp-team.com",mt=threads)
odd=converttoyuy2().selectodd().separatefields().selectodd()
interleave(even,odd)
weave()
converttoyv12()

Yeah I was going to ask the same thing. If you could re-up the missing parts ... or make a torrent on the spleen.

Here's an improved script with no need to do any sharpening now. It throws away the existing interpolated lines (the even lines) and starts again, quality is improved over the version I posted above. If you wanted you could put LSFmod it back in with a conservative value for strength in the range of 30-60 or so.

The GrainRestore setting improves the sharpness, so if you boost it from 0.7 to say 1.0 it will sharpen the image also. Personally I'd leave it on about 0.7 and use LSFmod with a setting of about 40 to bring the sharpness up a bit if that's what you want - but I don't think it needs to be quite as sharp as the source (which is due in part to the sharpness of the aliasing).

orig=avisource("great_white_clip.avi").killaudio()
orig.ChubbyRain2(th=7,sft=15)
converttoyuy2()
separatefields()
assumeframebased()
selectodd()
nnedi3(dh=true)
stackvertical(crop(0,1,0,0),orig.converttoyuy2().crop(0,479,0,0))
converttoyv12()
QTGMC(Preset="Placebo",TR2=0,NoiseProcess=2,NoiseRestore=0,GrainRestore=0.7).selecteven()
centre=crop(10,0,-10,-2)
centre=centre.BlindDeHalo3(rx=3,ry=2.5,strength=150,sharpness=0.5,tweaker=0)
stackhorizontal(crop(0,0,10,0), \
    stackvertical(centre,crop(10,478,-10,0)), \
    crop(710,0,0,0))
#stackhorizontal(orig,last)

Quick notes: Retaining the video grain (GrainRestore setting in QTGMC) won't do anything for you when authoring for DVD since it'll all be blurred out by the MPEG2 compression, but it will look nice if you ever decide to release it as MPEG4 AVC. You don't want the video looking "processed".

Screenshot of this script...

Before:

After:

By the way, I posted this in another thread, if you had the desire to make it 60i you could just add this to the end of my script and you'll get a full proper interlaced version (a very close approximation to what the original 60i video would have looked like):

even=last.converttoyuy2().separatefields().selecteven()
threads=10
super_params="{pel:2,gpu:0}"
analyse_params="""{block:{w:8,h:8,overlap:3},
    main:{search:{coarse:{distance:-10}}},
    refine:[{thsad:200}]}"""
smoothfps_params="{algo:21}"
super = SVSuper(super_params)
vectors = SVAnalyse(super, analyse_params)
SVSmoothFps(super, vectors, smoothfps_params,url="www.svp-team.com",mt=threads)
odd=converttoyuy2().selectodd().separatefields().selectodd()
interleave(even,odd)
weave()

Probably not a good idea for release, but not a bad one if you want to watch it on a CRT tellie.

Okay, so the de-rainbowing setting I used was chubbyrain2(th=3,sft=20) - it seemed to get rid of all the rainbowing in that clip, but of course you'd have to look over the whole file to find out if you need to use even more aggressive numbers than that (and I'd suggest to tune it so you don't use more aggressive numbers than you need since it'll make the chroma look flatter). I don't think you're going to be able to do much about that chroma bleeding - you could have the affected scenes professionally re-coloured, but I don't think there'll be much you can do about it in Avisynth. On the plus side there's not much bleeding in the scene you provided.

Here's a quick script, see what you think. It takes care of the haloing and aliasing (from the deinterlacing) as best as I can. You can't go too agressive on the dehaloing or you'll make a visible haze around the hard edges. You can use inputtype=2 in QTGMC if you want however the anti-alising will work better with the interlaced mode on and there should be a negligible difference in quality (it's negligible in the sample anyway).

#NOTE: Much better script in next post!

orig=avisource("great_white_clip.avi").killaudio()
orig.ChubbyRain2(th=3,sft=20)
QTGMC(Preset="Placebo",TR2=0,NoiseProcess=2,NoiseRestore=0,GrainRestore=0.5).selecteven()
centre=crop(10,0,-10,-2)
centre=centre.BlindDeHalo3(rx=3,ry=2.5,strength=120,sharpness=0.5,tweaker=0)
centre=centre.LSFmod(defaults="slow",strength=120)
stackhorizontal(crop(0,0,10,0), \
    stackvertical(centre,crop(10,478,-10,0)), \
    crop(710,0,0,0))

# Use this line to compare to original
# stackhorizontal(orig,last)

Good luck!

Before:

After:

towne32 said:

Right, I understand the assumptions. But if it turns out to be true: Is it possible to encode something such that the max bitrate is differently specified for different segments of the same file? Perhaps much fancier encoding software is required (than, say, adobe)? BDInfo spat out the numbers chapter by chapter, and there were no noticeable dips at branching points or in the short m2ts files. But of course if it's only 11 seconds with lower bitrate it would be masked by the average. 

I'm mainly asking because if there is indeed a 20mbps max for the entire film, it might actually be preferable to not use branching at all. Assuming one is only creating two versions of Star Wars, for instance, and has no qualms going from BD25 to BD50, a higher bitrate could be used without branching. Obviously, that goes out the window if 3+ branching options, or if the disc was already intended to be a BD50 packed with bonus material. A very inefficient use of space, but it should look better.

Well yes if you were limited to 20Mbps you could just set the 2-hour film to CBR and be done with it (you'd still need branching to fit two versions on though). I suggest testing my theory by trimming 5.825 seconds from the ends of the braching-points of the videos, and then joining all the heads together in one file, and all the tails together in another file and checking the average bitrate. Of course if it's lower than 20Mbps it doesn't prove anything since there's a 68% chance the sample would fall within 1 SDs of the mean, and 95% chance that the sample would fall within 2 SDs of the mean, however if it's higher than 20Mbps it proves the theory wrong (at least for the heads or for the tails of the joining points). And you could continue testing the other discs if it was lower than 20.

No interlacing shouldn't matter. Did you make sure to put converttoyv12() in your script? Many filters including chubbyrain2 and gradfun2db/mod are only compatible with YV12 and don't work even with YUV (which is the interleaved equivalent and often the colourspace supplied to Avisynth). If you want to keep 100% quality (i.e. the full luma quality) you could try doing this:

pointresize(width(),height()*2)
converttoyv12()
chubbyrain2() #or gradfun2dbmod()
converttoyv16() #or converttoyuy2()
pointresize(width(),height()/2)

# (edited to fix the linebreaks)

Also, I suggest using AvsPmod instead of VDubMod which is obsolete now. For the credits if they're all black-and-white you could just convert that section to greyscale using greyscale().

Did you try chubbyrain2?

Intruder said:

Okay, thanks for the notice. Would you recommend leaving SetMemoryMax out and let AviSynth control it?

Sorry let me clarify. The SetMemoryMax does set the caching amount for each clip in the script, but I'm not sure exactly how it's allocated and whether they are all loaded into memory at the same time or if some clips become unloaded to make memory for additional clips. As far as I'm aware the memory allocation is dynamic and it accommodates additional clips by reducing available physical memory for each one automatically, so yes I would leave it alone.

Yes I think you can have a higher average so long as the "overlap" parts (each end of each clip) is within the set amount. And I should note I don't know that it's 11 seconds or half of that or double that either, since we're just assuming it pre-loads the clip 11 seconds in advance.

Intruder said:

Infognition also recommends multi-threading: http://www.infognition.com/tutorials/rip_guides/avisynth.html

Just a quick writeup of your v10 with multi-threading, only three lines added hopefully make the deal. I have not tested it.

SetMemoryMax(2048) #I have enough memory

That's an incorrect use of SetMemoryMax and will cause problems. SetMemoryMax only controls the caching amount for each clip within the script, as Avisynth is 32bit it can't index beyond the 2GB limit imposed on 32bit applications (and that includes all 32 bit processes). Each time a new clip is created in the script more caching memory is allocated, your instruction could result in Avisynth trying to index around 22GB of RAM in total in DrDre's script.

Wow found a duplicate of my torrent. Really?? I don't mind it being shared on public trackers, but what's the point in renaming the file first?! GRRRR.

Also, great work on this one zombie84. :)

MrBrown said:

RU.08 said:

The English release has the subtitles instead of the burned in text for "The machines rose from the ashes of the nuclear fire ..." as noted here. It is missing the original mono soundtrack, and only has the 5.1 remix. It's a physically different transfer to the one used on the USA/Canada MGM release being that it is a PAL transfer and the US/CA one is a NTSC transfer.

 So it is the same, as the spanish and german. I think it will also be the same with the french release from 2001.

Yes without the extra audio, but with a whole array of subtitle languages.

The English release has the subtitles instead of the burned in text for "The machines rose from the ashes of the nuclear fire ..." as noted here. It is missing the original mono soundtrack, and only has the 5.1 remix. It's a physically different transfer to the one used on the USA/Canada MGM release being that it is a PAL transfer and the US/CA one is a NTSC transfer.

This is the DVD he means (or a localised version of it):

http://www.amazon.co.uk/Terminator-Special-DVD-Arnold-Schwarzenegger/dp/B000056IFJ/ref=sr_1_7?s=dvd&ie=UTF8&qid=1438429181&sr=1-7&keywords=terminator

"Special Edition" with the silver coloured cover. Dirt cheap to get a used one.

I actually thought it looked great overall in the cinema on 35mm. But that's not to say that it was the full quality of a film originally shot on 35mm, and it was a long time ago now. I didn't see the DLP, and I've never seen a DLP of that era. I saw the new Mission Impossible film the other day on DCP on a 30 foot screen and it looked great. Note that it was shot on 35mm. Skyfall also looked great, and it was shot entirely in digital at a resolution of 2880x1620 cropped to around 2880x1225 or so (2.35:1). In fact that resolution of 2880x1620 should theoretically be very similar to the resolution of 35mm film, depending of course on the quality of the film stock and how fine its grain is.

The majority of the original effects could potentially been rendered in at least HD res, however this was for a 2002 release and so possibly some scenes could have been rendered at a higher resolution than others. Toy Story 1 was originally rendered at 1536 x 922, but in 2009 it was re-rendered at a higher resolution for re-release (at least 2K), including being rendered in 3D.

The Star Wars effects are obviously much more detailed than the "plastic" Toy Story effects, but I imagine it's a similar process that's probably even more streamlined, to re-render the film at a higher output resolution. While I can only speculate, I would think that the film transfer was done at a higher res than the DLP one, however some of the shots at least were probably not at the full resolution used by the majority of the 35mm version. I guess the only way to tell for certain is to get a release print and examine it!

Yes I remember the effect where Padme rolls out of the container in the droid factory looking pretty bad (and it still does), and when the clones are fighting the Jedi in the dust cloud it looked shocking (and that shot looks improved now on DVD/HDTV/BD).

AOTC was filmed digitally but released on both DLP and 35mm projectors. Some people here suggested that the CGI was rendered at a higher resolution than 2002 DLPs could display (which is 1280x1024 with a 1.9:1 anamorphic lens). Note that the filmed elements were filmed at 1980x1080 with a 1:1 pixel aspect ratio and 3:1:1 subsampling (horizontal luma subsampling of 1440 pixels and horizontal chroma of 480 pixels). Essentially that's 1440x1080, and that was then cropped to 1440x800. But the release resolution for DLPs was 1280x1080.

No I haven't. Just an update, this coming week I will have the time to work out how to get my script to encode without running out of memory. The major problem is that there are too many different clips that get loaded all into one Avisynth file, therefore I'm pretty sure the problem will be solved by delineating the iterations into separate files - there are a couple of issues though, for one the DVD-upscale part of the script was never intended to be done in this way, so I need to rewrite it so that it only returns the required frames and then rewrite the other part of the script so instead of seeking out the required frames it trims from the returned file.

Just a quick note, I've changed my mind about the 1080p version. I will instead provide a playlist file that can be used with a rip off the Skynet BD to produce the result.

Here's the download link (2kb):

https://mega.co.nz/#!6FNEDTLQ!nFomkRz0-4QZOy74A_FCrPswKGvczAz8_TOaz3TaEXQ

Note that it is just the playlist file and you need a copy of T2 Skynet to use it.