HD "pop" rule of thumb

[MichaelTiemann]

I have noticed a fairly consistent pattern across a wide range of footage I've shot with my AF100 and HMC150 cameras when the subjects are human or human-scale. What I have observed is that when shooting wide or medium-wide shots, the subject does not deliver the expected HD "pop" on the screen. However, when shooting closeup, the "pop" is there. My theory is that as humans, we have a lot of experience being this >< close to people and things, and when we're that close, we can see a lot of detail (especially with our fovias). When an HD camera is able to display as good or better than what we see with our fovia, the image "pops" and we instantly recognize that we're looking at great image quality. I've run some tests on human-scale objects around the studio, and what I've found is this (for my AF100 camera):

When the distance (in feet) * 4 = focal length in mm, we are at the threshold of HD, but no pop
When the distance (in feet) * 5 = focal length in mm, we are at the threshold of the pop
When the distance (in feet) * 6 = focal length in mm, we get a definitely "hd pop"

For example, looking at an "interesting" sport coat with lots of texture in its weave and fabric, a 35mm lens (6 ft * 6 = 36mm) gives me an image that really speaks HD. When I zoom out to 24mm at the same distance, I can pixel-peep the image on screen and say "yes, the resolution is there, but no pop", and when I zoom out to 14mm, the resolution goes away and I think to myself "who would ever wear such a jacket on camera?"

I'm sure this is not a new discovery, but curious what similar rules of thumb might be out there. For example, having read that the C300 delivers 100% of the 1080 lines of vertical resolution (compared to the AF100s measured 680 lines), that would suggest that the C300 could provide the same pop 1.588x further away. Now, because of a different sensor size, the equivalent length in mm also changes by 1.25x, so the rule would predict that my jacket would deliver a decisive pop at a distance of 9.5 feet with an objective of 45mm, or alternatively, at 6' with an objective of 28mm.

 

[David W. Jones]

I have been shooting for over 35 years, and yet have no idea what you are talking about!

 

[Patryk_Rebisz]

Everything is relative.

You go on a date with some boring girl and she keeps on telling you boring stories and you are bored, but then you go on a date with a boring chick who's not event saying anything and you are bored to tears - you would love to talk to the boring talkative girl at the time. So is the case in your scenario. You get a medium shot with a guy in a jacket that makes it decently interesting but when you zoom out the "decently interesting" is competing with a cool table in the corner, a crowd of people in the background, over active branch of leaves in the foreground... This is where cinematography - real cinematography - comes in. How to drown out all the other voices so the "decently interesting" is still main point of viewer's attention.

BTW this relativism is part of many aspects of cinematography (life too, but it's another story). For instance most people assume 7D to be sharp mostly because most of the footage shot with that camera is mediums and CUs so you have a large head of the actor standing against some blurry background - people see the eyes, see the eyebrows, all in relations to blurry BG and they say "wow, how sharp!" but when you ever see a wide shot shot with this camera (where you can't compare what's sharp against what's blurry) you know you are back in the SD land...

 

[j1clark@ucsd.edu]

I have noticed a fairly consistent pattern across a wide range of footage I've shot with my AF100 and HMC150 cameras when the subjects are human or human-scale. What I have observed is that when shooting wide or medium-wide shots, t

I'm sure this is not a new discovery, but curious what similar rules of thumb might be out there. For example, having read that the C300 delivers 100% of the 1080 lines of vertical resolution (compared to the AF100s measured 680 lines), that would suggest that the C300 could provide the same pop 1.588x further away. Now, because of a different sensor size, the equivalent length in mm also changes by 1.25x, so the rule would predict that my jacket would deliver a decisive pop at a distance of 9.5 feet with an objective of 45mm, or alternatively, at 6' with an objective of 28mm.

The AF100 when using the internal compression engine, results in 4:2:0 color sampling. The C300 from the look of it, supports 4:2:2. In the 4:2:0 case one gets 1/2 resolution for B, R colors, and could 'explain' your impression there is more 'pop' for the 4:2:2 image at smaller object sizes on the resulting HD image.

I'm going to interpret the word 'pop' to be 'looks sharper', or 'more detailed'... and that's because... the 4:2:2 sampling has more information...

 

[MichaelTiemann]

Patryk,

You are correct that one should not rely on image quality alone to make up for a lack of story, character, emotion, etc. Watching educational videos do, however, cause one to start to look around the frame, and when the frame is gorgeous, that's better than when the frame looks so-so.

j1clark: my tests were with the SDI output (4:2:2) directly feeding an HD monitor, so no 4:2:0 problems there.

Again, the "pop" that I'm talking about is that once the camera is sufficiently close to the subject, the revelation of details make it absolutely clear that we're watching in HD. When the camera pulls back, of course one can still see more details than lower def images, but that theoretical pixel advantage does not translate in terms of visual impact they way it does when closer in.

 

[Amr Rahmy]

cam you provide a couple of images as an example to better illustrate your point.

 

[Justin Kuhn]

yes...it's easier to make a close-up look good than a wide, particularly with the cameras in that price range. So?

 

[maranfilms]

Did you have a couple pops before making this thread? kidding

I know what your saying, certain focal lengths produce better looking footage. I think patryks post makes a lot of sense as to why this is.

 

[Donald Ong]

Yep, my head just 'popped' reading that (joke). But hey, it actually makes sense

 

[Finnegan]

Let's start this discussion by defining just what the hell "HD pop" is, because I have no idea what you're talking about. And I'm not the only one, evidently.

 

[morgan_moore]

I think there are two things?

Simple hard framing.

Compression in complex scenes leading to a wooly image?

 

[Donald Ong]

Let's start this discussion by defining just what the hell "HD pop" is, because I have no idea what you're talking about. And I'm not the only one, evidently.

I think he's referring to how close-ups generally look sharper than wides.
For example, shoot a 200mm close-up on a 7D, everything looks nice and dandy, shoot the same scene with a 17mm, uh. Soft images + moire everywhere. Since he's using the af100 and hmc150, neither of which can out-resolve most DSLRs by much, he's likely facing the same problem. Of course us laymen haven't come up with a term for this phenomenon Apparently it's called 'HD pop". heh

 

[Tolin]

Get a hobby

 

[j1clark@ucsd.edu]

Again, the "pop" that I'm talking about is that once the camera is sufficiently close to the subject, the revelation of details make it absolutely clear that we're watching in HD. When the camera pulls back, of course one can still see more details than lower def images, but that theoretical pixel advantage does not translate in terms of visual impact they way it does when closer in.

I don't think this is any 'news'... Film film had the same penchant... in fact, stills shot under strobes often look 'sharper' than images shot under continuous for the same camera settings... Perhaps due to harder light, less small motion because of the shorter duration of the strobe... perhaps the 'color' of the light... all add up to a 'different' feel of the image.

Humans are more sensitive to 'edge' characteristics in images, and so a 'close up' will have more edges. But even then often it centers on 'what edge'. For the human face, the eye lashes are a critical element. If those are 'high contrast', the overall impression is that the image is 'sharp'... even if stuff has been 'fuzzed out' with selective retouching... And of course humans are much more 'detail oriented' when it comes to human faces... so, as a frame is more filled with a face there is more 'detail' acquired...

However, even with all that... a wide for an HD image should look 'better' than the equivalent SD... otherwise something is wrong.

 

[MichaelTiemann]

[...]

However, even with all that... a wide for an HD image should look 'better' than the equivalent SD... otherwise something is wrong.

I'm not saying it doesn't. All I'm saying is that to my eyes, merely being able to resolve line pairs on a monitor is not nearly enough for the image to really pop on the screen. It takes many pixels speaking together before there is convincing detail (as opposed to intelligible noise). My question to this list is whether this really is just a function of pixel counts on the viewing monitor (in which case a higher-resolution C300 will not significantly beat a lower-resolution AF100) or whether this does have to do more with the capture resolution, in which case a C300 will allow 50% more working distance to achieve the same perceived level of detail. I kinda believe the latter is true, in which case the C300 would be an absolutely decisive winner compared to the AF100 for instructional videos that require mostly medium shots.