Tru but, maybe dis duznt make sense…dun unwant da sample to be good like you wanna hear??
Yeah I measure this sheeyat then correct using this guys genius method to make sure what goes to my ear holes is along the Harman curve.
Setup basically those in ear mics, F3 as interface, USB out to REW which runs the sweeps.
Its not too complicated. I only convinced one guy to do this on audio science board. The rest are possibly spec fetishists and not interested in actual outcomes I guess.
1 Like
This the chart I mean to post earlier. Its average between left and right but you get it. This show the harman target for reference
Lol the stealths on target down to 10hz 
This has all the target curves you could want to align to using method above.
Learning about kinoscopes. I been informing how I restore vids based on NTSC specs from the time but should be targeting kinoscope specs cuz that what survive and not the original broadcast info sadly. This how I approached the original cziff improv vid and it improvement better than say tatum or moiseweitch vids of late.
Test cast for going deep there gonna be Ogdon’s Dante sonata.
I was thinking about this.
Best sources are DVD at 29.97FPS (deinterlaced from 60fps half fields)
Which is a 3:2 pull down capture of a 24FPS Kinoscope
That Kinoscope is a 24fps snapshot of a 60FPS half frame image pulling down at 30.
What we are seeing is a far cry from what these broadcasts actually looked like on the TV humming along at a clean 30fps form the 60 interlaced.
I think if someone just transferred at 24FPS in high resolution the kinscope films it would be DRAMATICALLY better. The monitors they used in in the kinscopes were 525 lines (with some now for picture) and DVD itself 480. If there an algo to throw away half frame scans and interpolate motion blur also could be improved a lot.
Hmm…
1 Like
I apply new learning to Tatum files I had already.
This is the source with stabilization of frame only:
Here my attempt at fixing pull down artifacts, and cleaning up a billion other things plus upscale
Fatty’s Learnings for Make Better Glorious Old Piano Vids
directed by and starring da Fat Mofo
1 Like
Yeah flow changed a little
- Stabilize
- De interlace like DVD to progresive
- Interpolate frames like a kinoscope source
- Redo pulldown with clean progressiveness instead of half rolled BS.
- Upscale in Topaz with goal of avoiding weird artifacts
- Clean in neatvideo
- Fix weird ass brightness variances and white / black balnce in Davinci
Finito!
I don’t think the AI will know what to do with her nostrils.
2 Likes
da tru test will be to ztabilize da MART channel
1 Like
Ooh! You got any super short original vids that haven’t been raped by youtube into VBR compressed turds? Can see what it does.
Hoping for extra ears and fingers.
Wait I think I got an old Yuja bee. Will noodle and embed here. The ogdong is gonna take a couple days but I haven’t started cooking it yet.
1 Like
Three hours left on this one minute video if it don’t error out. 
3 hrs for a one minute video = 315 hrs for a 105 minute movie…just a tad over 1 week.
1 Like
Yeah doing a whole movie poss melt my computer. The moiseiwitsch took a few days. Might be diff when back in seattle tho. Had an RTX something or other in the home machine. This taking a super long while cuz I gotta do frame interp to fix some weird color blurring.
1 Like
It errored out trying to do higher framerate. Meh… here is the 25fps upscale tho. Imagine it twice as smooth and with less chroma smear on her arm flapping.
mart production truly adverzarial
1 Like
Anybody have connections at UCLA? They have minty version of this stuff in their Philip Kahgan collection. The tube vids basically turned into blobs when I tried doing stuff with it cuz google is woke, I believe.
Proof very HQ transfers exist.
Actually they probably dun need tinkering with. Someone with a real channel should put that stuff up.
