Last edited by Lunchbox; 2007 December 10th at 13:35.
Since nobody is replying here, I'm answering my own post
The footage from the Downsizing looks a lot better than the footage captured from the DV Lock.
Both of them shows some vertical strip lines on the image. I don't know how to get rid of them.
I wish mine looked as good Taky. I didn't see any vertical lines on your clips on my computer (run in VLC and Creative). The only thing I noticed was some jaggies on the groom's shoulder in the dvlock footage.
By the way....beautiful the cat. Unfortunately my two would tear apart the turtle in about 2 minutes.
I saw those vertical lines when I blow up the image to full screen in my Dell 30" LCD screen.
My cat is very mellow. I can stack a bunch of stuff on her and she wouldn't care.
As for any interlacing artifacts, I don't notice any on SD DVD's. Computer files however do need to be deinterlaced.
Check out my videos here: http://www.youtube.com/24Peter
Using 25P and using DV lock didn't show any difference between that downsized with Premiere Pro 2. Maybe certain NLEs downcovert better? I am torn between going toward Vegas instead of upgrading to CS3, if the HD editing in Vegas is leaps and bounds better that Premiere this will sway me.
It shows noticeably difference in downsizing for me. To use DV-lock in the cam, it is downsizing in realtime whereas doing it in NLE gives more time (slower than realtime). It takes longer time but it yeilds better quality.
There are different ways to deinterlace in Premiere. I tried
- Deinterlace during export
- apply "Always deinterlace" in the timeline
- apply "Interlace consecutive frames" in the timeline
- apply "flicker removal" in the timeline
I compared all the clips and the deinterlacing during export yeilds the best quality.
OK, I did an export to AVI with deinterlacing and downsizing and the image quality improved ever so slightly. Honestly, I could be very happy with the in-camera conversion if I am working with a long clip.
The difference between the original HD material and the in-camera converted SD material is noticable, but the difference between the two downconverted versions isn't that obvious to me.
What is impressing me the most is the quality of the resulting clips in either HD or converted SD. I was thinking about getting am XM2 for higher quality SD stuff but now I don't think it is necessary and I would rather spend a little more and get the XH A1.
The difference is noticeable to me. I did a frame capture and you can see it side by side. Check out their face and the groom's shoulder.
Last edited by Lunchbox; 2007 December 10th at 13:35.
Yes, I can tell the difference...but still compared to other SD cameras it still looks awesome and DV lock is quicker especially with long footage.
What I did is to convert HDV to DV footage with a laptop. It won't occupy my main computer.
There are processes in converting, say like, the final video to MPEG-2. There are 1-pass and 2-pass encoding. I wouldn't mind doing 2-pass for the quality.
If you are outputing for web or youtube, don't bother
In the viewable pics you posted the bottom one has more definition just buy looking at the shingles of the roof.
Maintaining the highest definition for SD DVD will look much better on the BIG screens. Try adding a bit of sharpness in Premiere without getting "edging" effects.
I got excellent results with 60i footage by using avisynth to first separate the fields (into half-height 60fps images), resizing with Lancoz4Resize which gives a bit of sharpening and then weaving the results back into 30fps interlaced footage for DVD. Like so...
aud=MPASource("footage MPA PID 814 DELAY 3ms.mpa").DelayAudio(.003)
When I tried to resize directly in various encoders, I got lots of interlacing artifacts. This method yielded nice looking SD footage.
(You may be wondering about the audio delay - a tool called DGIndex pulls the mpa audio out of the clip and provides the correct delay to apply to the audio).
Great thread, good info.
Recently I've been letting the camera do all the downconverting (DV-Locked). The resultant capture looks pretty good, but I'm always interested in extracting better quality video.
The main problem I'm having right now is when exporting to DVD. The image quality is way underwhelming... it looks like VHS. I went so far as to play some of my DVDs on friends TVs/DVD players (just to rule out my TV/DVD) but, as noted, we were less than happy w/ the results. What's strange is that when you play the DVD on a computer, it looks a little better.
I'm going to experiment with external downconversions. Taky, if you don't mind, can you describe to me how you capture and downconvert? I'm assuming that I can capture m2t via HDVSplit, then downsize it to DV-AVI w/ StreamClip. From there, load into Elements, cut it & export to DVD.
Is it possible that my burner is at fault? Will a higher end burner make more of a difference? Any/all comments appreciated.
I have a similar problem. After I viewed all the nice footage from HV20, I recently noticed the video from my GL2 sucks bad! I used to think they are top quality video.
I don't think the burner is the problem. The problem might be the MPEG2 encoding. If possible, try to select 2-pass encoding instead of one.
What I did was to capture m2t using HDVSplit. Load into Premiere and export to AVI while selecting the Deinterlace option.
I would love to select 2 pass, and adjust the quality settings, but Premiere Elements doesn't allow me to tweak any of the "export to DVD" settings. Unless you're referring to the export to MPEG2 (or AVI?)... which I will certainly attempt.
While the footage is on my computer, it looks great... as soon as I export to DVD, the quality takes a noticeable nose dive.
I use Canopus Procoder to encoder AVI to MPEG-2. For some reason, the MainConcept encoder built within Premiere doesn't seem to do a great job in MPEG2 encoding.
Sorry to bring this thread back, but being in a 25p/50i country, I have a question for Taky (or anyone else) which is probably well known to those of you in 24p/60i countries...
If that's 24p, where is the conventional 3-2 pulldown? On a DVD (or VHS, or broadcast etc) you get 3 fields of one frame, 2 fields of the next, and so on. This means some _video_ frames consist two different _film_ frames stitched together.
In your files, all the frames are pure progressive frames. There's no interlacing. What you get are 3 unique frames, and then a 4th that's duplicated. So you get frame numbers 1,2,3,4,4,5,6,7,8,8 etc. 4 original frames spread over 5 resulting frames. 24*(5/4)=30, so it's a frame repeat pattern (rather than field repeat) to get 24p content up to ~30fps.
Is this what NTSC HV20s output? Or is my DV decoder being strange? In felt sure this must be the case, but I see the same in VLC, media player classic, virtual dub, and TMPGenc - IIRC they don't all use the same DV codec.
Sorry to bring up an old thread, but I just got an HV20 and have been doing all the DV vs. HDV tests.
I can confirm that HDV to SD converted in software is *slightly* better than HDV down-sampled to DV in-camera on-the-fly with "DV Lock" output.
BUT...in my tests, standard DV recorded in-camera (setting the HV20 to record in standard DV mode) also seems to be slightly better than HDV down-sampled to DV with "DV Lock". Which would put it on a roughly equal level with HDV converted to SD in software...but takes a lot less time.
I am assuming this is because it's the "on-the-fly / real-time" MPEG-2 transcoding that the HV20 has to do when down-sampling that introduces the (very) slight quality loss. Of course, real-time transcoding cannot be as good as a method where the computer has time to analyze each frame at its leisure.
But with native DV mode recording in the HV20, there is no "live" re-compression needed. The HV20 is actually recording a standard definition DV signal to tape in the first place.
Of course, with this method you don't have an HDV master copy to archive, but if this doesn't matter it would seem one could save oneself a lot of re-rendering time for good looking standard-definition projects.
Last edited by ral-clan; 2011 July 2nd at 14:17.
If you shoot in DV mode your are recording with uncompressed audio using an I-frame only video codec. The advantage of uncompressed audio is obvious. The advantage of an I-frame only codec is that it records independently encoded views of the same scene over time. Taken together these independently encoded views contain significantly more spatial resolution than any single frame on its own. Since HDV uses motion prediction to create dependencies between frames, then DV obtained by down sampling HDV does not have the same frame encoding independence as native DV recording. However, I'm surprised the difference is noticeable.
Last edited by ejolson; 2011 July 3rd at 02:03.
Also note that DV and HDV have the same bit rate on tape. So, obviously, more is lost when compressing a HDV 1920 image, than a DV 720 image - even considering the different codecs...
In actual use, though, I find the diffrence so minute that I always record HDV - then I have a HD original if I ever need it. (Upscaling DV to HD looks much worse...)