Thursday, April 15, 2010

HS10 vs cropped dSLR megazoom … how close?

Disclaimer: This test is for distant shots … the kind that require 720mm :-) … close-ups tend to have larger details and thus are not pounded as much by the famous Fuji selective noise reduction.

Yes, they are at it again. Unseasoned people at the FTF seem obsessed with the HS10 as dSLR-lite. Tastes great! they say. Less filling! counter those that know a bit more about what works and what does not. Cropping works when pixel quality is high.

Yet another comparison thread was started recently, the specific topic of which is to complain that the comparisons so far were not fair because the dSLR lenses that were cropped were not of the mega-zoom variety and thus were not the kind of lens that would be a reasonable choice instead of the HS10 and its rather comical speed-of-light X zoom lens. He was, of course, referring to sunshine_boy’s crops.

Three quotes from his post follow … they show his biases quite clearly … and frankly, I’m not impressed. But I like the challenge, so what the heck.

So.... I would love to see someone try and match the HS10 with a DSLR output from a Canon, Nikon or third party 18-200 mm lens- a truer comparision[sic].

It would be of interest to see if a DSLR could best the HS10 with such a lens..but I am thinking it aint[sic] gunnah[sic] happen ..sooooo

You DSLR fan boys prove me wrong !!!!

So I thought I’d show some full-sized samples from around the net, processed with suitable copyright acknowledgement of course … and I claim fair use for educational purposes if someone finds this objectionable.

I will then display similar images from the 18-200VR and post those, cropped to reach 720mm effective.

Now … how much to crop? Well, if we crop away about 38% of the frame, we’ve basically taken the field of view from 450mm to 720mm. So images from the 70-300VR at full zoom are candidates. Images from the 18-200VR at full zoom require more cropping than that, about 59%.

In this latter case, if we end up with about 41% of the original linear dimensions, we must multiply that together to get the area that is left and we come up with about 17% … or about 2mp. More than enough for display on the web.

So on with the show.

I’ve not been impressed by either the sharpness or the contrast seen in the 720mm images from the HS10 … the images tend also to show edge artifacts and smudging of smooth areas. Watch for this as you proceed through the examples. The dSLR cleanliness makes a huge difference. I compare against my D300 with the 18-200VR, the first generation one, so the HS10 has a real advantage being 3 or 4 years newer as a design.

Image 1:

The first image is one I found on focus-numerique.com and is a building turret, shot at full zoom. As always, click through to see the 800px versions.

HS10_1[1]

The closest I could come in my archives was a shot of the Parliament buildings in London from across the Thames. Here is the original image.

d300_200_1[1]

And now the 720mm crop:

d300_200_1_cropped[1]

The difference in the low contrast details -- edges between adjacent bricks for example – makes this image a lot crisper than the 720mm image from the HS10.

Some will be thinking, but there are 5 times as many pixels in the HS10 image (10mp vs 2.1mp) so I can print larger. Mt reply is HUH? The image is mushy at 0.1mp on the web … it will be a tragedy printed any bigger than 4x6.

So I think we’ve already established the difference between a tiny sensor with mushy pixels versus a serious camera with very clean pixels. I don’t really have to go on … but I’m going to pound this point home to make sure that the “dudes” on the FTF get it.

For sh*ts and giggles, let’s add another lens to the mix juist this once. I used to own a manual focus 105mm f2.5 lens that was unbelievably sharp. I got rid of it because it was too difficult to focus reliably. Still, I got some stunners from it.

d300_105_1[1]

That looks sharp even at this size. What happens when we zoom in to 720mm? To do that we need to chop all but 0.5mp … that is, this file has 1/20th the number of pixels as the HS10 file.

d300_105_cropped[1]

I think I’d rather put that on the web or on a 4x6 print any day of the week. Clean pixels and sharp lenses … why the heck do you all think we buy these cams in the first place?

Image 2:

There aren’t all that many images out there of buildings. But I found one on Picasa … the details are fairly clean on this one, much more so than the previous one. However the contrast sucks. Sorry, but I looked all over to find something better.

HS10_2[1]

The image I matched it with was also shot on a crappy day … heavy cloud. Many will recognize the London city hall.

d300_200_2[1]

And the 720mm crop:

d300_200_2_cropped[1]

This isn’t exactly a crushing victory, but the crop is competitive and certainly has far better contrast. Still, we cannot really compare that since the two were shot in somewhat different light. There is a crispness in the crop that is missing with the HS10, but it does show good detail in that shot, so we can call this a draw.

The third shot is a long lens image of people walking. I’ve seen numerous images of people with this cam and those at 720mm share a lot of similarities … mushy smooth areas, helmet hair, very fuzzy edges to clothing, even CA. The examples on the review at photography blog have some too.

HS10_3[1]

To isolate some of the nastier issues, those that make the image useless at anything but web sizes, I put together some crops.

HS10_3B[1]

Well, that’s not pretty. Her glasses are not to obad, as they are high contrast. But her hair and ear are total mush. The branches have terrible edges. Her pants and shoes have heavy CA … looks like Fuji is correcting it in cam somewhat, but the tell-tale edge artifacts remain to rob sharpness. Looks pretty awful to me.

I had to search a long time to find a random shot ot a person that would almost fill the frame *after* cropping … I rarely shoot people as it is. But I found one … just one. A woman jogging through one of the parks in London.

d300_200_3[1]

There is really no contest on this one. Rathyer a slaughter in my opinion. Confirmed by the crop:"

d300_200_3_cropped[1]

So there you have it. The 720mm lens is no doubt to real hoot to play with, but just like the S8000 when it first came out, the actual results are not anywhere near what was hoped for. Simply mediocre pixel quality.

I’d shoot the 18-200VR and crop any day of the week instead of using that lens. And remember that there is a consumer lens in use by many that goes to 300mm. Had I used this lens instead, the crops would have been 5mp … and that’s more than enough for a perfect 8x10.

Bottom line … dSLRs like the D300 and its sensor cousins the D90 and D5000 easily dust the HS10 with the consumer grade megazoom 18-200VR after cropping to match.

Sorry … but that’s just how it is.

5 comments:

Unknown said...

This is not a fair comparison. At least you need to show some critical information such as ISO information, lighting condition, shutter speed etc. DSLR with cropping can be surely better but you need to tell us the difference in price to get the difference in image. For the cost of D300 and 18-200VR, I’m pretty sure Fuji can make something better than HS10. Isn’t 24mm still wider? The difference is minimal and I bet 18-200VR is better lens but often people pay much more for the small difference.

Kim Letkeman said...

Daniel: I was answering a simple assertion that a dSLR with kit lens could not match the HS10 long zoom. I answered it by showing that it can. Nothing more, nothing less. If I ever get the chance, I will perform a head to head test, but for now I was only trying to answer the assertion I had read to my own satisfaction.

Jan said...

Thank you for this comparison. :-)
I sincerely hope you will get the chance to do a more scientific test.

I myself am torn between 550D+Tamron 18-270mm and HS10 for this very reason... cropping a 18mp picture with finer pixels and better sensor... sounds so tempting.
How do you thing this lens would fare for this?? :-)

Since I never owned a dslr type cam, I will go for HS10 now to get the hang of shooting in PASM and to see if I can stand to carry the cam around and survive. :-)
HS10 is cheaper entry for my experiment, and it is a leap forward in IQ in low light situations and high ISO usability compared to formere P&S's

And yes as a rookie i am afraid of lack of Scene Recognition mode on dslr's :-P

Kim Letkeman said...

Jan: In general, I think a cropped Tamron image shot correctly (i.e. stopped down when you can, and you almost always can) would be superior to the HS10 at 720mm. Basically, the same number of higher quality pixels on the subject.

The HS10 is a fun camera and does afford the ability to learn PASM controls and external flash (the hard way, since Fuji do not offer a TTL flash solution.) I have never used scene modes for my shooting, preferring to control the exposure myself.

I wouldn't call the HS10 a leap forward ... the sensor in there is the same as that in the SOny TX1 and WX1, which is very nice at high ISO, but not that far ahead of the Canons (G11, S90) and the Fujis (F200EXR, F70EXR in medium res.) However, you will definitely get nice images at night if you take your time and try to get good sharpness.

I too hope that I get a chance to perform a more scientific test of the HS10 against a dSLR for cropping. I would enjoy that :-)

Jan said...

Thank you very much for you answer and your POV. :-)