Hi folks,
This generally happens either at the very top of the clip or the very bottom, and on flat areas: There are brief moments where some light pixelation appears and then goes away. I raised the lumagain to 4. My max bitrate is 9800 with a 9500 minimum, it's not bitrate starvation (although possibly raising the minimum to 9600 may help??).
The other factor I'm considering is if the possibility of my AQ being too low, perhaps? I set it at either 0 or 1. But I read that AQ actually helps with these flat areas if I raise it?? (kinda confused on this one). Thought raising AQ PREVENTS extra bits from going to these flat areas, thus creating some pixelation.
Any help is appreciated.
Thx!
Try StreamFab Downloader and download from Netflix, Amazon, Youtube! Or Try DVDFab and copy Blu-rays! or rip iTunes movies!
Closed Thread
Results 1 to 30 of 45
Thread
-
-
Is the source progressive or interlaced? Try setting one or the other in the settings instead of allowing it to default.
Also, try lumgaim = 1, and AQ =2 or 3.
Your bitrate seems a bit restrictive, have you tried making it wider ? Then set a high average.
Do you want it to be DVD compliant?
Is your source material very demanding? Have you tried CQ mode?
-
The source is 29.97fps telecine NTSC (it's 80's animation). So it's not entirely interlaced. Should I set it to interlaced anyway?
Lumgain 1 is fine, except those flat, dark areas need extra bitrate to avoid the pixelation.
Can you explain AQ to me a bit? If I raise it, will it help with flat areas or restrict bitrate?
I've considered making my bitrate wider, yes. The thing is, the actual source was grainy. I cleaned this video up before sending it to HcEnc and am trying to use the highest possible average and maximum bitrate in order to obtain the highest visual quality with no artifacts.
Yes, I need it DVD compliant, which is why I'm setting it to 9800 the highest.
HcEnc does not do CQ very well, I have noticed (and also read)
-
Isn't 9800 the maximum allowed for video + audio? Is your movie silent?
What is your source resolution and format?
Regarding AQ, flat area's with little detail can be starved of bitrate, raising the AQ number allows more bits .
Is your source particularly difficult? Perhaps you could post a few screen shots showing the trouble,
or even a short clip of the source.
Assuming your source is real telecine (and not field merged or something harmful like that)
leave the progressive/interlaced as default (do not specify them).
-
I only use encoders in conjunction with DVD converters and MPEG editing software, so I can't advise you on matrices etc,. however it seems to me that the way you want to use HCEnc is not the way it is supposed to be used.
HCEnc can do a fine job with one-pass VBR encoding and two-pass VBR encoding, but from what others have said about it, HCEnc is not a CBR encoder. If you intend on making the upper and lower values for bitrate very close, that is pretty much what a CBR encoder does. If you want CBR, use something other than HCEnc. Maybe QEnc would suit you.
If as davexnet said 9800 is the maximum bitrate allowed for video + audio together, the bitrate you are using is too high for DVD. Since you stated in a previous thread that Windows DVD Maker did a perfect job, why not look at its output to see what the maximum bit rate was. It used 8000 kbps max bitrate for my test disc from last year. I think it uses CBR encoding, because the maximum and minimum bitrates were very close.Last edited by usually_quiet; 7th Nov 2011 at 15:02. Reason: correction
-
Dave,
My source is Telecined wit fields. Is this a problem? What d you mean "harmful"? Most NTSC shows are Telecined with fields. I'm confused about your "harmful" statement.
Re AQ: Ok, so you say that raising AQ DOES increase bitrates in flat areas, correct?
usually_quiet: I know, I know (regarding Windows). I feel like a moron making the statements I did in the past about Windows DVD Maker and HcEnc...and would like to put all that nonsense to rest. After doing ALOT of experimenting I can't believe I even compared the two. Trying to move on here....moving on:
I've gotten so used to HcEnc, I really woudn't want to use Qenc. I perfer to set the average and maximum bitrates close to each other (to simulate CBR) than go to another encoder. But I'm not sure if that's a good idea. I mean, is it? Would it make a difference? I mean, common sense tells me that although I may be using more bitrates that I need (hence inflating my file size), I would certainly not be ruining my visual quality. And I don't care about the file size, really I just want good quality.
There's another encoder that I heard about that is used strictly for anime I think, but can't remember the name now. Does it ring a bell to you?
-
The shows as broadcast, or the DVDs made from those shows? Yes, the broadcast versions are as you said. When on DVD some are 'hard telecined' (the telecine encoded into the video - 2 of every 5 frames interlaced, although anime is different unless you find a passage with movement every frame). Many are also 'soft telecined' (encoded as progressive with 3:2 pulldown). Much better for encoder efficiency would be to perform the IVTC in your script to return it to progressive 23.976fps and then encode it as such, with pulldown applied.
I perfer to set the average and maximum bitrates close to each other (to simulate CBR) than go to another encoder. But I'm not sure if that's a good idea. I mean, is it?
-
"When on DVD some are 'hard telecined'. Many are also 'soft telecined' (encoded as progressive with 3:2 pulldown). Much better for encoder efficiency would be to perform the IVTC in your script to return it to progressive 23.976fps and then encode it as such, with pulldown applied."
What is the best way to find out what I'm working with here? Would g-spot tell me?
"...perform the IVTC in your scriptperform the IVTC in your script"
What is a good script to properly do this? Actually if I could do it in Virtualdub it would be better. But if Avisynth does it better, please post script.
-
Given the presets I have seen AVStoDVD use, I suspect that the algorithms HCEnc uses for VBR encoding do not work very well when the upper and lower limits are too close.
The rule of thumb I have followed for my HDTV to DVD conversions is to specify a lower limit that is half of the upper limit. That has typically been 4500 kbps and 9000 kbps. (9000 kbps is the highest maximum that AVStoDVD permits.) manono knows his stuff so I guess I could go even lower.
I don't know what combinations you have already tested, but you could try some significantly smaller lower limit values on a short test clip and see how it works, and if all is well, move on to encoding larger segments.
-
unclescoob - by "harmful", I meant that sometimes, interlaced/telecined files are incorrectly handled (for example, resized).
If you captured them yourself or similar they should be fine.
Take a look at the HCenc BIAS parameter. It's used with 2-pass and allows you to limit the vbr "spread".
eg. If you set min=1000, max=9800, avg=8400, bias=100, it will *almost* be CBR @ 8400.
I doubt that bitrate starvation is the cause of your issue, something else may be going on.
-
Yes, it will.
If it says 23.976 Pics/s and 29.97 Frames/s and 59.940 Fields/s, then it was encoded (DVD, right?) as progressive with pulldown. If 29.97 Pics/s and 29.97 Frames/s and 59.94 Fields/s, it's hard telecined (unless it's something more peculiar, such as a bad PAL to NTSC conversion).
If the former you make the D2V using Forced Film; if the latter with Honor Puldown Flags. I'm assuming you're using DGIndex to create a D2V file.
You can also find out what you have by opening the VOB(s) in DGIndex and running the Preview. Or by playing a VOB and pausing so you can advance a frame at a time (with the deinterlacer turned off)
What is a good script to properly do this? Actually if I could do it in Virtualdub it would be better. But if Avisynth does it better, please post script.
TFM()
TDecimate()
You can get the TIVTC.dll and its docs here:
http://avisynth.org/warpenterprises/
I agree, not when the bitrates are as high as they are. It could easily be something as simple as an improperly calibrated monitor and/or TV set.Last edited by manono; 7th Nov 2011 at 18:39.
-
The anime DVDs I have were either purchased or recorded using a DVD recorder, so I can't tell you which MPEG2 encoder is favored by anime fans.
The best I could do is try to remember more freeware encoders I have run into. The only other freeware MPEG2 encoders I can recall having used (but not without another program controlling them) were ffmpeg, mpeg2enc and maybe MEncoder. By most accounts QEnc and HCEnc are better.
-
If it says 23.976 Pics/s and 29.97 Frames/s and 59.940 Fields/s, then it was encoded (DVD, right?) as progressive with pulldown. If 29.97 Pics/s and 29.97 Frames/s and 59.94 Fields/s, it's hard telecined (unless it's something more peculiar, such as a bad PAL to NTSC conversion).
If the former you make the D2V using Forced Film; if the latter with Honor Puldown Flags. I'm assuming you're using DGIndex to create a D2V file.
You can also find out what you have by opening the VOB(s) in DGIndex and running the Preview. Or by playing a VOB and pausing so you can advance a frame at a time (with the deinterlacer turned off)
What is a good script to properly do this? Actually if I could do it in Virtualdub it would be better. But if Avisynth does it better, please post script.
TFM()
TDecimate()
You can get the TIVTC.dll and its docs here:
http://avisynth.org/warpenterprises/
I agree, not when the bitrates are as high as they are. It could easily be something as simple as an improperly calibrated monitor and/or TV set.[/QUOTE]Last edited by unclescoob; 8th Nov 2011 at 09:44.
-
Manono,
My source is a 29.97fps NTSC retail DVD. You're suggesting for me to IVTC, therefore bringing it back to it's original 24fps. Then dumping it in HcEnc, and adding the pulldown flags, bringing it back to 29.97fps.
Here's my question: Why would I IVTC and then re-do it again in HcEnc? Does this allow it to encode my video better? And if so, what happens to the fields in the process? Wouldn't I have to deinterlace as well?
-
IVTC with pulldown flags allows you to encode progressive and at 23.976 fps, rather than interlaced at 29.97. Progressive MPEG2 encoding is superior to interlaced MPEG 2 encoding. And the lower frame rate spreads the available bitrate over fewer frames for more bits per frame. Pulldown flags are just a few bytes per frame that tell the DVD player how to produce 59.94 interlaced fields per second from the progressive frames.
A hard telecined source has at least two fields from each film frame (unless it's been edited after hard telecine, in which case you sometimes have orphaned fields). IVTC figures out which pairs of fields to weave together to restore the original progressive film frames.
In addition, a soft telecined DVD allows a DVD player to send the picture to the TV as progressive frames. Upscaling progressive frames gives superior results to upscaling interlaced frames.Last edited by jagabo; 8th Nov 2011 at 10:03.
-
Jagabo,
I'm just going to briefly break down what you said (and present you my workflow) just to make sure that I understand this correctly:
Telecine is the process of slowing down (progressive film) from 24fps to 23.976ps (via pulldown).
IVTC is the reverse - speeding back up 23.976fps back to 24fps (progressive).
Here's my workflow:
1. IVTC my AVI clip via Avisynth (which will bring it back to 24fps (progressive).
2. Dump my IVTC'ed AVI clip into HcEnc (at this point it is 24fps Progressive).
3. Click on the "pulldown flags" tab in HcEnc (this will then "pull down" to 23.976fps).
4. Encode
So far, so good?
-
No. Telecine is the process of producing 59.94 fields per second video (TELEvision) from 24 fps film frames (CINEma). It usually involves slowing the film down to 23.976 and applying 3:2 pulldown.
IVTC is the process of turning 59.94 field per second video back into 23.976 progressive film frames.
Note that 59.94 fields per second video is usually stored as 29.94 interlaced frames per second when digitized.Last edited by jagabo; 8th Nov 2011 at 11:02.
-
Yes. Assuming your AVI is 29.97 fps from telecined film. Note that the progressive output from TFM().TDecimate() will be 23.976 fps. Also be aware that pulldown flags can be used for any progressive frame rate from 19.98 to 29.97 fps. So you may occasionally run across a video for which 23.976 isn't the correct rate.
-
Yes, my AVI is 29.97fps from telecined film. This AVI was produced by Virtualdub after having cleaned up my VOB clip using the program. Now I am ready to encode back for DVD.
-
Be careful when filtering interlaced video in VirtualDub. It does not handle interlaced YV12 chroma properly. The chroma channels will end getting blended together. It's more noticeable with animated material because of the large motions and highly saturated colors:
https://forum.videohelp.com/threads/333462-Need-advice-VHS-capturing-device?p=2068357&v...=1#post2068357
It's best to convert to YUY2 or RGB before giving the video to VirtualDub. For some sources forcing the input color space to YUY2 works in VirtualDub's Video -> Color Depth dialog.
-
YV12 is native of NTSC video, correct? Is this why HcEnc does not allow your video input unless it's YV12 first?
Also, I convert to YUY2 in the Color Depth section at Virtualdub to avoid the chroma problems. So I don't need to convert it first with Avisynth before sending it to Virtualdub. My video looks great.
-
Analog NTSC video has the chroma channels encoded at lower bandwidth, and hence resolution, than the luma channel. The most similar digital format is YUV 4:2:2 (YUY2 and similar formats) and NTSC DV's 4:1:1.
MPEG2 encoders usually use YUV 4:2:0 (YV12) internally. Hence the YV12 requirement of HcEnc.
That doesn't always work. It works when opening MPG files directly in VirtualDub. It doesn't work when opening interlaced YV12 files with AviSynth. You must explicitly convert to YUY2 or RGB in that case.
-
That's what I mean. First I open my Mpg (ripped VOB actually) file in Virtualdub, change the color space in "Color Depth", do all my cleaning, and export the AVI. I then make an AVS file and convert to YV12 before sending it to HcEnc. Of course after what I just learned, I will also add IVTC to that script chain before sending it to HC.
Any thoughts?
-
You're saying your 'workflow' as detailed in the post just above this is the best and most efficient way to do your filtering? Is there some reason you don't use AviSynth for the filtering? Is there some VDub-only filter you absolutely have to use that's not available in AviSynth?
And you're creating a lossless AVI, right? Why not create it for YV12 already? I can see creating a lossless AVI and do it all the time when my filter chain is slow (and the only reason to use VDub here, in my opinion), but all filtering (including the IVTC) can almost certainly be done in AviSynth on the way to making that AVI for HCEnc. And if the filter chain is reasonably fast, just feed the AviSynth script into HCEnc.
Maybe your workflow can be fine-tuned if you let us know the filters you choose to use in VDub, along with providing a sample of the source.
-
In a short answer: I like Virtualdub!
It's an excellent tool if used correctly. I can see my adjustments in real time, instead of having to reload scripts, and retype arguments and/or settings, to then continuously re-load. Virtualdub is a fantastic tool and it's never given me any problems.
I understand that Avisynth can assist with accomplishing various tasks. I've used it for purposes other than frameserving, and it's a great tool too! However, my only goal is to clean up my video from dirt and grain, in order to make it look presentable on TV (I am aware that I can do this with Avisynth--please keep reading). I use Neatvideo, which is an excellent plug-in for Virtualdub, as you might be aware of.
I understand that I can call Neatvideo in AviSynth. However, considering my ONLY goal is to clean video and nothing more, I don't see the need to take the extra step. The interlaced chroma issue is easily handled in Virtualdub if I directly load my MPEG file in it and adjust in Color Depth, and that's all there is to it. I have done many tests and my video, after filtering with V-dub, looks fantastic. And I don't mean "oh it's good enough, let me be lazy and leave it at best" fantastic. These are videos that I care for and carefully look for every little detail, stitch and flaw. No doubt, Avisynth is serving it's purpose with frameserving to HC, converting to YUV12 and now, IVTC. So I don't completely dismiss it's purpose for my goals. But if V-dub is doing a great job with cleaning up my video, again, why take unecessary extra steps?
So I guess I can return the question back to you, Manono: Now that you know what I specifically do with Virtualdub and my results, is there some reason why I shouldn't use it? For cleaning purposes, what would be so different if I did it with Avisynth? Am I messing up my video by using both, Virtualdub and AviSynth? (a powerful combination, as expressed by many). I'm now curious.
I don't know whether or not the following are or aren't available in AviSynth, but I like to use Denoise by the fIsh (based on Donald Graft's filter: Deflicker). Not to mention others like temporalcleaner, temporalsmoother (with reasonable settings), DNR and of course, Neatvideo goes without saying. I'm sure I can call these filters in Avisynth. But again, if I can use it in Virtualdub, why go through the extra steps?Last edited by unclescoob; 9th Nov 2011 at 09:44.
-
Well, I don't (and won't) use NeatVideo, so if that one's a necessity you'll have to do as you say - either use VDub or use it in an AviSynth script (which doesn't help much as it'll still require you to convert to RGB and back). All the rest have their (better) AviSynth equivalents and don't require the colorspace conversion. Some of the better AviSynth denoisers/cleaners are slow so the idea of making an intermediary lossless AVI is a good one. If you're married to NeatVideo, then you have no choice but to do it your way, I don't think. But if you'd put up a small 10 second sample as suggested by both me and jagabo, we might be able to suggest other ways to achieve your goal.
This pixelation, it only becomes visible in the completed MPEG-2 out of HCEnc, and is nowhere to be seen in the lossless AVI you feed into HCEnc? Now that you've detailed your filter chain is it possible some of your 'pixelation' (posterization, really, maybe?) is caused by one or another of the filters you're using? Have you tried encoding without the filters, or doing test encodes while at the same time pulling out one filter at a time to possibly try and track down the culprit? One of the greatest mistakes people sometimes make is to over-filter but maybe you're not using the ones you mentioned all at the same time.
Similar Threads
-
Audio correction in VirtualDub - possible for selected periods of video ?
By ralf07 in forum AudioReplies: 5Last Post: 18th Oct 2011, 23:10 -
Problem with HCenc 0.24.
By HolyFries in forum Newbie / General discussionsReplies: 0Last Post: 17th Jun 2010, 14:18 -
HCenc Question
By Maikeru-sama in forum Video ConversionReplies: 6Last Post: 18th Aug 2008, 13:01 -
HCEnc vs MVBR
By awesomer in forum Video ConversionReplies: 7Last Post: 31st Jul 2008, 19:12 -
HCEnc
By Belfour in forum DVD RippingReplies: 1Last Post: 28th Jun 2007, 10:57