![]() |
Gaps in capture histogram from procamp sharpness?
2 Attachment(s)
I recently got my baseline levels for my ATI 600 USB procamp settings to a good place where it safely captures most tapes. But while browsing the forums I found multiple suggestions that the sharpness should be set to zero (I had mine at default of 2), so I tried a few captures with this. I quickly realized that setting the sharpness to 0 was adding gaps into my capture histogram.
From my understanding, gaps are bad and can be causing banding or posterization. I hopped into Virtualdub to modify my procamp live to confirm it is the sharpness. What I found is that with any Sharpness other than zero, there are no gaps. There is a very small difference between 2 and 1, 1 creating some slightly deeper valleys. Increasing sharpness beyond 2 incrementally smooths out the valleys (and spikes), cranking all the way up 15 makes everything completely smooth. I played with the other settings to see if anything else would cause gaps, I only found increasing the contrast (well beyond my normal baseline) would also create gaps, but I don't think I would ever be capturing at those levels. So, I am stuck between conflicting information: on one hand I have read that I should always reduce the sharpness to zero; on the other hand, setting to zero always produces gaps in my histogram. My intuition tells me to take sharpness to 1 but I would appreciate any recommendations or explanations. Attached are two screenshots of the histogram: Sharpness = 2 Attachment 19084 Sharpness = 0 Attachment 19085 |
Let me illustrate with an analogy. Imagine the most beautiful woman you've ever seen. Intelligent, fun. Cooks, cleans. Good with money. Comes from a wonderful family. Body count of only one. And she adores you.
Me = "Marry me!" :hearts: Puritan = "Body count? Tramp! Whore! Stone her to death!" :eek: __________ If everything else looks good, then don't worry about a sharpness 1-2 value? What you have to realize is that, at a very nuanced level (which 1-2 is), analog is not 100% intrinsically "sharp" or "unsharp". It's analog. It's demodulated and processed. Yes, you can give it "artificial sharpening", but you can also give it "artificial blurring". The key is to know which is which, and most people do not. Most of the quality (non-junk) capture cards are defaulted to the most ideal values, or just a hair above (added sharpness). Sometimes fully disabling values can have unwanted side effects, as if the case here. For the ATI 600 specifically, there's a good post at VH (and my reply) from a few years ago: Sample image https://imgsli.com/MjAxMDE3 Quote:
Don't be a puritan (purist). Those people are nuts. :screwy: |
Thanks LS, that was the direction I was leaning towards. I guess I was wondering why there have been multiple threads specifically recommending disabling sharpness on the ATI 600 USB if it causes this. I am going to stick with 1 or 2.
|
3 Attachment(s)
For anyone who might come across this later with similar issues, I wanted to add the Sharpness is only affecting the luma. I did also notice later on U and V also had gaps in their histograms, but the sharpness had no effect on them. Instead, I had to return the saturation control back down to the default of 32.
I will be keeping my procamp settings at default from now on, except for brightness/contrast as needed to avoid illegal values. I did notice one other oddity, in some areas I see extreme spikes, almost like the reverse of the gaps, extending all the way up. I don't mean at Y<16 or Y>235, these happen between 16-235. I am not going to obsess over getting perfect captures, but I am a curious person, and I just want to understand what this is. I did try a capture with the procamp settings all at default and still produced the spikes, which I have read elsewhere on the forum if this happens it is unavoidable with that card. Also, I am using the levels histogram in AVI here, so it's been converted to YV16 and I am wondering if the spikes are just coming from the conversion actually. I don't see the same kind of spikes in the Virtualdub histogram. If I understand correctly, the gaps are usually caused by rounding errors when the capture card procamp is adjusting the signal. Are the spikes caused by a similar issue? I guess that a gap is missing data, so a spike would be a clustering of a lot of data into a single point. Does this just mean there are a lot of points at the same luma/chroma value in that particular scene? The next question is, I have read the gaps can cause banding/posterization, can spikes also cause artifacts that I should be looking out for? I am not going to go too crazy over this, just curious. Below is an example of the spikes, you might also notice the gaps in UV (later fixed with Saturation setting), and values bleeding into Y>235 (I increased the contrast a little as I decided I wanted to capture in 16-254, and use levels to come back to 16-235). I did wonder if running levels before the conversion would clean up the spikes with something like this: Code:
Levels(16,1.0,255,16,235,dither=true,coring=false)Attachment 19099 Attachment 19100 Attachment 19101 |
Site design, images and content © 2002-2026 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2026 Jelsoft Enterprises Ltd.