The Myth...

S

Shaw

Veteran
Hey guys,

Lately I have been very bothered by the idea that downconverted material looks "sharper" and "better" than footage captured at the native resolution. This just doesn't seem plausible at all with regard to the math. You can't pull more detail out of a set amount of pixels..... am I missing something? It seems like it's one of the biggest myths around unless someone can prove otherwise (please do!).

I've been around DVinfo.net and you constantly hear people talk about how much "better" downconverted HDV is compared to DV... I'm just not buying it....

Help?
 
Re: The Myth...

What do you mean, "more detail"? HDV's resolution is much higher than DV's, and if you're downcoverting, you're losing detail.

If you were to take footage on a VHS camcorder and play the tape, it would not look nearly as good as something recorded in DV and then recorded to VHS. The same idea applies to DV and HDV, though it's not exactly the same in practice.
 
Re: The Myth...

Not sure what I mean. What I'm getting at though is why do people claim to see a difference in downconverted material? I can't seem to find a logical explanation for it... What is the math behind this?
 
Re: The Myth...

When you start with HDV, you start with a much higher resolution than DV, and much more detail information. There's simply a lot more there to start with. When you downconvert, you lose some of it, but it still looks better than something which was recorded at the lower resolution.
 
Re: The Myth...

Right :)

I understand that but I can't find any evidence that that is true. Afterall you only have a set amount of pixels... you can't resolve anything extra...
 
Re: The Myth...

It's going to be 720x480, but like the difference between progressive and interlaced, it's not something I can describe to you with words. You have to see it.
 
Re: The Myth...

quite right. The thing is I can describe the difference between interlaced and progressive in terms of human perception but I can't with downconversion... there has to be an explanation I just cant seem to find it :(
 
Re: The Myth...

Well, you can talk about the persistence of vision all day, but when someone asks you to describe what progressive looks like and how it's different from interlaced, there are no words which will give that person an accurate mental image.

Like I said, DV on VHS tape looks much better than something recorded by a VHS camera, but the DV is "downconverted" to VHS "format," so theoretically, it should be the same, right?  But it's not.  When you have much better quality to start with, the end result is better.
 
Re: The Myth...

True better in creates better out but I would think that such highly compressed material would make for worse DV. Anyway, I'm going to have to run through the math and see whats going on.
 
Re: The Myth...

Try this. Take a 5MP photo and a 1MP photo of the same thing. Reduce the size of the 5MP photo until it's the same size ad the 1MP photo. See which one looks better.
 
Re: The Myth...

Will do. Except I'm not sure I have a 1mp camera... hmm. I'll have to take a look around. If anyone here gets an FX1 could you do some down converted comparisons with the DVX (and deinterlaced I suppose to keep the motion the same)?
 
Re: The Myth...

Well its hard to make any conclusion between 1megapixel and a 5 megapixel because they have other feutures that would have effects on the picture. Like the lens.
 
Re: The Myth...

Here's an idea to toss into the mix...

condensing sharpness results in increasing acutance.
 
Re: The Myth...

Great question Shaw... perhaps the great Barry G needs to chime in on this one... I'm sure there's a formula behind it :)
 
Re: The Myth...

davids right, it does make a difference, and ...i think barnes is on the right path here. but the issue where the confusion is stirred up is that people may be confusing references to mathematical pixel resolution with percieved optical resolution.

simply put you are correct a 720 by 720 image can only have so much resolution, but image sharpness equates to optical resolution. therefore an image at 720 x 480 which is not sharp  appears to have less res to our eyes than one that is sharp.   when you downconvert, you are also essentially sharpening  the image by condensing the resolution.  the result will be an image at the same resolution  .. but one which is much sharper and appears to be more resolved.   just as you can sharpen your native 720 x 480 images, and make them ..well ...sharper.. and percieveably better if sharpness if what you r going for..

to quickly understand how this works.. just think of the opposite factor.. when you blow somthing up. or expand an image.. it gets softer .. like projecting your DV onto a theatre screen.. its going to look softer because everything is spread out. ..like 28 days later.... but take the same image, and show it on a tv, and its looking pretty sharp and clear because the edge that was 4 inches wide on a big screen  is now only .02 inches wide, so its going to look sharp.     .. so when you take an HD res image with reg sharpnes, and shrink it to SD dimensions. it obviously doesnt have more pixel res, but  lines and edges and everything will appear sharper because it is condensed.  if it was a bit blurry before, it will looks sharp after.

a very easy test would be this .. take a HD res image  with a white background, and draw a single black line circle, or simple shape ..say 5  pixels wide, and blur it  with a gaussian blur a little untill it looks a little out of focus.  now just resample the image to SD dimensions, and notice how the bulr seems to be sharp again.. its too small to see because its condensed. ..just the same as being at wide angle on a lense makes everything look in focus.. because the things which are out of focus are optically resolved so much smaller.. that you cant tell its blurry. like looking at a big somewhat blurry picture from 40 feet away.. it looks sharp. so the images WILL likely appear to be sharper.  if your blurred at native res. your just blurred. if your blurred and then shrink it.. it looks sharper. if your sharp and then shrink it.. it just looks even sharper.

hope this helps.
 
Re: The Myth...

[quote author=MovieSwede link=board=Offtopic;num=1099615491;start=0#11 date=11/05/04 at 02:57:06]Well its hard to make any conclusion between 1megapixel and a 5 megapixel because they have other feutures that would have effects on the picture. Like the lens.[/quote]

Not so. Many 5MP cameras will also take photos at 1MP.
 
Re: The Myth...

iamloser: thanks for the post! If I understand then running a simple unsharp mask (much as photozoom does for part of its processing) will give exactly the same results as downconverted material?
 
Re: The Myth...

[quote author=Jim Brennan link=board=Offtopic;num=1099615491;start=0#14 date=11/05/04 at 07:14:29]ohhhhh, THAT myth.[/quote]

:D
 
Re: The Myth...

it might result in a similar visual effect, but the process is different.

An unsharp mask filter works by increasing the acutance between areas of high contrast. When viewed at a normal distance, it gives an increase in sharpness by simply defining edges.

If you look in close when you apply an unsharp mask filter, lets say the ends of spiky black hair against a grey background...you'll notice that an unsharp mask increases the contrast of the pixels immediately surrounding the areas of tonal difference. Essentially, you’d see like a very thin white halo around each chunk of hair black hair that when viewed at a normal distance, the localized increase in contrast translates visually as an increased acutance.

Now, while you can apply an unsharp mask to a low resolution picture and increase contrast, applying it to a high resolution picture results in a much more precise effect. Having more pixels to work with allows you to really refine the level of sharpness in the effect. Also, in lower resolution pictures, you wind up sharpening the inherent noise as much as the actual picture.

Reducing a high res image to a low res one has some of the same visual effect, but you’re really just averaging pixels rather then affecting sharpness…even though you see sharpness as a byproduct.

As imloser said, if you take a 100 pixel fuzzy highlight and reduce it to 40 pixels, the result is a much sharper highlight, but that’s not the same effect as applying an unsharp mask to the same highlight.

At least...that's how I think it works.
 
You must log in or register to reply here.
Back
Top