Neat video vs denoiser
DVD-video is fixed capacity and has limitations on max video bitrate. Note you can't use a limitless number for bitrates. So the problem in your case is both a lack of noise, and bitrate allocation They are applied preferentially to problematic areas, not a general type of noise. That is essentially what dithering filters do. In order to "convince" that encoder that those areas are important, you need to add noise to those areas, thus lower quantizers are assigned.
Since the bitrates are lowered in those regions, you get macroblocking (when the bitrate isn't sufficient, you get macroblocking). Raising quantizers means lower quality, lower bitrates. You're seeing what is being fed into the encoder.įlat and dark areas are deemed not as important to most encoders, so higher quantizers are assigned. Similarly when you look at the output pane in vdub, you're looking at decompressed image with filters applied. The encoder doesn't "see" or detect bitrate it sees the decompressed image. Low complexity video (denoised, very smooth) requires lower bitrates to maintain a certain level of "quality". So during encoding, the encoder detects low bitrates and raises the quantization to make up for lost detais.hence the macroblocks. The low bitrate statement that Smurf said made sense, as when you denoise, you're removing a part of the video, perse (even if it is noise, it's still a part of the video). The macroblocking that I'm referring to specifically appears AFTER I apply denoise filters. Poison, yes I'm referring to macroblocking that I see in dark and/or flat areas. But you need higher bitrates or better compression for this to work effectively, often it's not possible with DVD MPEG2 compression But to human eyes, a fine dither is barely noticeable.
gradfun2dbmod, this "convinces" the encoder to allocate more bitrate to those areas that it usually drops, because it "sees" those areas as noise and lower quantizers are allocated. You can add specialized fine noise in gradients and dark areas by dithering e.g. x264 (AVC encoder) excels at this, but I gather you are making a DVD, hence stuck with MPEG2 video. You can use adaptive quantization to allocate more bitrate to dark areas. Unless you're using a very poor denoiser that generates block artifacts even before compression. Then deblock or cpu=6 wouldn't help you in that specific case, because from what you've said it's a function of encoding. If the macroblocking occurs only during encoding, then it's not in the source. Using a higher bitrate can help, but I gather you're more interested in a specific scenario, like macroblocking in dark scenes, not general macroblocking Blocky artifacts are quite normal for that.
Yes technology has improved, but I gather from your other posts that you're making a DVD. You should continue in your own thread instead of hijacking others, because there is relevant information you have omitted Adding CUP(6) for example during postprocessing can help. Has this changed? Is it possible to eliminate noise in video without blocky artifacts from forming during encoding? Or rather, is there a special script that can be composed in Avisynth that can prevent this? I know this post is old as I'm sure the technology has changed a bit from 5 years ago.
#Neat video vs denoiser Pc#
UNLESS- if you do NOT see it when you preview your capture on pc monitor, then it might be something else and I would have no idea.
#Neat video vs denoiser tv#
The noise on the bottom is hard to see in your attached image, but I think it is ordinary head-switching noise, you don't see in on TV viewing because it falls into the "overscan" region, and is only seen when you view on a PC monitor. It might soften your video more than you like, though. I'm surprised no one mentions Convolution3d. I don't care how slow, I'm happiest with this one so far of all I have tried. If you use the setting to denoise all 4 planes (Y,U,V,Luma) with a single parameter setting, it is quite fast, but if you have different parameters for each, it is deadly sloooowwwww. I second the vote for FFT3d, it's doing a really good job on my current project. You'll have to check the documentation for the particular filter you are using. Some of them have the default set to Interlaced = false, so they won't work properly on interlaced material until you change the default parameter. With many of the aviSynth filters, you must specify a parameter for Interlaced = true or = false.