I have been playing around with converting some episodes on DVD to avi files and I have been using AutoGK to do it. My question concerns resolution and overall resulting video quality. Here is a sample scenario... Let's say that I want to convert a 24 min. episode to a resulting avi that is about 300-350mb in size... If I were to reduce the resolution to say 512x387 instead of having it at 640x480 would this increase the quality of the video at all, or does the resolution compared to overall file size have nothing to do with the resulting video quality?
+ Reply to Thread
Results 1 to 18 of 18
A smaller frame size allows for lower bitrate.
512x387=198144 pixels, 640x480=307200 pixels.
The same quality encoding (bits/pixel) would in the former case be achieved with 0.65 x bitrate in the latter case. However, when you reduce resolution, you inevitably lose some detail.
The higher the resolution, the higher the quality. Unless the bitrate is too low, then resizing down will give higher quality.
Whether in your case 512x384 or 640x480 would give the better quality. I couldn't say though.
Well, you're using AutoGK for the job. If you let it choose the resolution, it'll do it based on the results of the compression test, and give you the highest resolution for the quality it aims for. If you screw things up by setting a fixed width, then there's no telling how it's going to turn out. If, for example, AutoGK determines that 512x384 (not 387) is the best resolution for the file size you're allowing, and you override that and choose a fixed width of 640, you can almost bet that it's not going to look very good, in spite of the steps AutoGK takes to try and improve it (softer resizer, lower bitrate quantization matrix).
...does the resolution compared to overall file size have nothing to do with the resulting video quality?
On the contrary, resolution with a fixed file size has everything to do with the resulting video quality.
Edit: And as guns1inger says, if the resolution gets too low, then the video will appear quite soft. In that case you have to increase the file size (and/or give it poorer quality audio, MP3 instead of AC3, for example).
As a general rule of thumb, if I want decent quality on a 4x3 video, I use a minimum of 10 MB per minute (24min X 10 Mb = 240 MB ). Sometimes this gives very good quality, other times it could be better. For your 24 minute episodes, 300-350 Mb should give you very good quality.
So like the others said, just let AutoGk do the work. My guess is that it should give you 640x480 nearly everytime based on the size of your files.
Thanks for the replies everyone!
As I said in my previous post, I am playing around with this program a bit just to see the different results that I can get and of course, seeing what will work best for me. You know what's interesting though... I encoded a couple of different shows that are both about 24 min. in length. The first one, I set the resolution at "auto" so I let AutoGK do the work and I set a target file size range at 300mb. We'll call this example "A". For the other show, which I will call example "B" I messed around with the resolution and set it at 512 width just to see the results I would get. I set the target file size on this to 275mb. Now, with "A" AutoGK encoded it with a resolution of 656x496.
Based on the responses in this thread, I would think that with the resulting avi files "A" would look better than "B" ultimately. However, in my eyes, that was not the case. I really thought that "B" looked much clearer and smoother than "A" did. In fact, in some of the high motion scenes in "A", it almost looked like there were some light artifacts. Any explanation for this? Also, why would AutoGK encode this at 656x496 instead of 640x480?
Thanks a lot for everyone's help!
Well, A has 65% more pixels/frame, but only 9% higher bitrate - hence greater risc for artefacts.
Did you view both videos at the same display size? Or did you view them both at their native size (for example, at 100 percent in WMP)?
When you shrink a video frame the results usually look clearer than the original. But if you enlarge the smaller one to match the larger you will see that it has lost a lot of detail.
For example here is a 320x240 image:
Shrinking it down to half (both dimensions) may make it look clearer:
But if you resize it back up to the original size it will be much less clear than the original:
If you were to watch both videos enlarged to full screen the larger would look much clearer.
Regarding bitrates, with Xvid, when you have insufficient bitrate you don't lose clarity, you get macroblocks. Here's the original image compressed too much with Xvid (this is way overdone to make it obvious):
As others have pointed out, the smaller your frame size, the less bitrate you need to maintain picture quality.
So, for a given bitrate (file size) it's a matter of balancing macroblocks against frame size.
When I encode I usually don't worry about file size. I set the frame size to whatever I want then encode in a constant quality mode (single pass, target quantizer mode in Xvid). I pick the quality I want and the codec uses whatever bitrate is necessary to maintain that quality. That way I always have exactly the right bitrate to get the quality I want.
Originally Posted by jagabo
Can you perhaps briefly explain what you said in your last paragraph, in terms of how you do it? Do you use AutoGK to do this? I am not sure I have ever done anything with the "frame size" at least not that I recall. Just from reading what you said, it sounds like you use a different program, but I could be wrong.
Edit: I should also point out that I am WAY confused when everyone is referring to bitrates in their responses. How are you able to tell what bitrate I am getting when I haven't given much info on the file that I am encoding? Also, how do you alter the bitrate?? The guide that I am using with AutoGK doesn't mention anything about bitrates nor do I see a place in the program that allows you to alter the bitrates that Xvid encodes with.
I use a tool called ASPECT to get the aspect ratio right,j ust manually crop the picture and it will tell you the correct aspect ratio of the movie on the dvd and the which resolutions you should use on your conversion to divx/xvid. For quality/size i follow what Gianni said 24min X 10 Mb = 240 MB ). or use Avi.net
Just thought of sharing this as i find this tool extremely usefull when finding out the aspect of the movie and which resolution to use on my divx conversion.
Ps: Another neat function is that it can make an avisynth script based on the cropping made manually, example:
1- rip your "main movie"from your dvd to your Pc and use Dgindex to creat a d2v script
2- open aspect, click on the Aspect ratio dropdown menu and select "advanced"
2- click on open and select "d2v project" and delect the d2v script, manually crop and click on ok.
3-back on main menu select autosize and it wiil find the best resolution, click on "save" and it wiil produce an avisynth script based on the cropping you made that can be "fed" on to virtualdub to convert to divx/xvid without going in to the filters section, just select the codec and you´re away.
Originally Posted by ASPECT homepageI love it when a plan comes together!
Originally Posted by TempestGarden
If you choose Target Quality, you will likely find your resulting file will be bigger or smaller than what you expected. For example, if you want to back up a 90 minute film to burn onto a 700 MB CD, you would not choose the target quality option, but would probably go for the Custom Size option and select something like 695-700 MB. But if you don't really care if a file is exactly 350 MB or 700 MB, then you could go for the target quality option.
I should also mention that when I said earlier that you should be getting 640x480 with your rips, that's because I usually set maximum width of 640 pixels for my rips when the source video is 4x3 (full screen). For widescreen sources, I use maximum of 720. This is so that I don't have issues with my standalone DVD player with its modest chipset. In any case you probably don't want it sized over 720 pixels width (DVD standard).
As to your question about the difference in quality in something like 640 vs. 512, AutoGk theoretically tries to choose the best size so that it will look best when viewing the file on a television set.
Originally Posted by TempestGarden
I usually use an audio bitrate of 128 CBR. This is generally the divx standard, and again I use CBR (constant bit rate) so as not to have issues with my standalone DVD player. If you are backing up a Music DVD then you might want to choose something higher like a 192 or 256 kbs bit rate.
You don't deal with the video bitrate in AutoGk, but inherently if you choose a bigger file size, AutoGk will be using a higher video bitrate than if you had chosen a smaller size.. The higher the bitrate (the larger the file size), the bigger the video frame can be without losing image quality.
There is a program called GSpot that will tell you what the audio and video bitrates of a given file are.
Thanks a lot for explaining this in a little more depth.
I ran into something else that I didn't think about before, that might be effecting the quality that I am getting as well. The Frames per Second of the video I am encoding. I am ripping tv episodes from a Japanese Region 2 DVD... Any idea if these are 23.97 fps or 29.97 fps as a standard? I am thinking that they are 23.97 fps but am not sure. When I use DVD Decrypter to rip the episode from the DVD, it does so with the resulting files being 29.97 fps. Just for giggles, I tried forcing AutoGK to change it to 23.97 fps by checking the "Reduce fps" option and the resulting AVI file was then 23.97 fps and still looked good.
Did I do the right thing here? If it was supposed to be 23.97 fps originally, why did DVD Decrypter rip it as 29.97 fps? Is there a program that I can use to check the actual DVD itself to see what the fps is supposed to be?
Originally Posted by TempestGarden
BTW this is what autoGK uses and it is very good at knowing if the original NTSC is really 29.970fps or if it is 23.976fps etc.
The only option you have control of is if a hybrid is detected. A hybrid will be 29.970fps at times and 23.976fps at times. With autoGK (using the hidden options panel) you can force a hybrid to either deinterlace (my suggestion) or to IVTC (I advise against this).
The option "reduce fps" you played around with is for a HDTV source. Don't play around with those options for a DVD rip or anything that is non HDTV. This is really for a true HDTV source like a TS file.
For myself I always force a width of 640 when I use autoGK since that is the "best" or "highest" compliant resolution you can use ... you just have to make sure you set the file size correctly so that you use enough bitrate to get a good quality image.
- John "FulciLives" Coleman
giannid hinted at this but didn't actually come out and say it, the relationship between file size and bitrate is simply:
file size = bitrate * running time
Of course you usually have both audio and video bitrates, and you have to use the right units.
Okay... another quick question.
I have been increasing the file size to get better results in my AVI since the bitrate also increases. The only odd thing is... No matter how high I am going with the file size, I am getting what looks like areas that are "alive" around the outlines on people, and oddly enough, I seem to notice them more when the actors are not moving as opposed to "action-type" movement scenes. I did de-interlace the video, which gave me better results, but I am not sure if there is anything I can do to get rid of this movement or "haze" that outlines the actors at times. Any suggestions?
Use a high resolution ... such as using a width of 640 ... also increase the file size to increase the bitrate.
Also MPEG-4 is imperfect and not as good as MPEG-2 unless you really MAX out the bitrate.
- John "FulciLives" Coleman
Originally Posted by FulciLives