I was wondering why Sony didn’t announce the new compact camera(s) yet. But finally we have some hint. Sony will announce the TX55 at any moment from now.
According to DCviews: “What sets the TX55 apart is the new digital zoom technology called Clear Image Zoom that, instead of cropping the image, is using a more sophisticated process that includes pixel creation and pattern matching for a clearer and sharper magnified image. The DSC-TX55 will come in four colors and cost around $350. The camera is expected to be available in September…”
Those are the specs
26mm wide-angle zoom
3.3-inch touchscreen LCD
10 frames per second (full resolution) burst mode
3D-still photos capability
dedicated video button
1080i full HD-resolution video capability
21 Responses
Finally, new rumor.. waiting all day.. hehehehe
Is this the only one that will be announced or just one part of a batch?
“instead of cropping the image, is using a more sophisticated process that includes pixel creation and pattern matching for a clearer and sharper magnified image”
If you hunt down other sources, then you find out this is interpolation using a fancy algorithms that attempts to reconstruct shapes using pattern matching of some sort. Of course it is still necessary to crop the image and it’s impossible to recreate the information content that wasn’t captured by the sensor. The best that can be done is to make a better informed guess – which is what this is about.
So a compact gets 1080i but the NEX-3C does not? Unfair!
Interpolation rather than cropping.
Aside from giving a higher output resolution than a crop, there is little to gain from it. You can’t create detail, no matter how advanced the algorithm is.
Obviously Andrea wants us to ask the questions “Does it work” and “Does it work well enough to use on an Alpha camera”.
+1!
It doesn’t. You can achieve some usable enlargement up to ~25% with best algorithms (and note: you cannot go with one algorithm! There are few best and you choose one depending on the content of picture, one never will solve many different problems unless you incorporate power of real CPU with image analysis) but beyond that you end with ugly artifacts.
There are no miracles, sorry. You can’t create something that doesn’t exist.
An Alpha camera can save Raw files, so you can crop and enlarge an uncompressed image in Photoshop or whatever software suits you.
This is a gimmick for JPG-only cameras.
I dunno.. interpolation and image enlargement can go a long way. It can’t add real detail that wasn’t there, but you can make canvas prints from an iphone with genuine fractals if the noise isn’t too bad
This is very important for sensor crop zooming in video mode!!!!!
Smooth extra 1-4x zoom.
Every time after image is cropped and before it’s encoded to 1080p, It should be carefully interpolated to keep consistent image quality.
Imagine how much zoom in video u get when You crop 24mpx to 2 mpx !!!!
Sounds like extra 4x zooming only using sensor crop.
Imagine that in video mode You will have perfect 16-220mm range out of 16-55mm lens. Retaining excellent quality in every step of zoom.
When this happens in A77 ofc:] (sorry guys, green tee betetr than coffe:])
Wow, sounds promising! I can’t wait to see how well this works. Any samples?
Just trying to imagine this through….. please don’t come down on me too hard.
What if the interpolation can suppress moire – so that it could replace a Bayer filter? That way, interp would allow for more actual information to reach the sensor….
The Bayer filter is not there to suppress moire – the Bayer filer is their to provide the colour rendition. It’s the job of the anti-aliasing filter to suppress moire. In effect the AA filter slightly blurs the image to just below that of the sensor. I don’t think that you can completely mimic the effect in software. It’s also nothing to do with interpolation which is essentially estimating what colour and brightness intermediate pixels will be when up-scaling images.
Nb. I should add that the AA only deals with moire pattern caused by interactions with the sensor spacing. If there are moire patterns caused my objects in the image (e.g. overlapping meshes and the like) it will have no effect. On very small sensors with very high resolution sensors AA filters are often unnecessary because the lenses can’t resolve down to the pixel size.
That’s interesting Steve. At what megapixel # is the AA filter not really necessary? I have read recently that some authors feel a 39 megapixel count will have the resolution to do some antialiasing that might combat moire, so that the blurring isn’t necessary. Perhaps the new FF Sonys will use that as a goal?
The point that moire starts appearing is determined by the optical frequency of detail as well as sensor resolution.
Things with very high frequency detail like fabric weaves, metal gratings and windows on skyscrapers from a distance still show it on high resolution sensors, whereas most normal photographic material will not show the effect.
It’s possible to low pass filter in post processing if there is any annoying moire in an image. A lot of photographers just don’t like to as it’s another step which takes time.
Sounds like there are always tradeoffs to be made. However, I am beginning to think that an interpolation function can do a couple of things that are important to high resolution photography. First, rather than adding a blurring through installation of an AA filter an interp function will fight the jaggies by filling in, rather than bluring edges.
Also, it will probably fill in the gaps between the pixel wells – and increase the signal to noise ratio and thereby decrease noise while filling the gaps and making an image seem more saturated.
I’m now hoping that Sony goes down the road of eliminating AA filters, and finding a way to substitute interpolation. Since most MF cameras don’t have AA filters, they have an edge in image quality that goes beyond pixel count, and like do some form of AA blurring in their image processing – but the key is that the processor is in charge of the amount done. Interpolation would/could/might be an interesting strategy to combat aliasing AND noise that would make Sony imagers much better for it.
At what point AA filters become unnecessary will depend on the resolving capability of the lens. In general, pretty well no camera lenses can achieve significant contrast (measured by the MTF) at 100 line-pairs per millimetre. That corresponds to a 5 micron pitch (as you need twice the sampling frequency). However, with a Bayer-pattern filter, we have a 2 x 2 cell so a 5 micron pitch on the repeating RGB pattern corresponds to a 2.5 micron photocell pitch.
If we translated a 2.5 micron pitch to a 24 x 36mm FF sensor that would come in at about 138MP (or about 61MP on an APS-C sensor). If we assume the rumours about the A77 are correct, and it has a 24MP APS-C sensor, then the photocell size will be about 4 microns. If this can be reduced by another 40% or so, then I think the AA filter can be dropped as I doubt any available lens will produce moire at an appreciable level.
The Sigma SD1 does not use a Bayer-pattern filter or AA-filter and has a 5 micron photosite pitch corresponding to 100 line-pairs per millimetre. I don’t think that’s any coincidence.
Of course the better quality the lens you buy, the more likely it is that moire will show up. That’s probably one of the key reasons that no main-stream DSLR manufacturer with a bayer-pattern has yet omitted the AA filter as they don’t want nasty looking moire patterns showing up on test charts using their best lenses. (Of course the Leica M9 does lack an AA filter, but they are, maybe, dealing with a customer base whose typical photographic subjects are less likely to exhibit moire).
Sounds interesting to me. I know past Sony cameras offered “smart zoom”. And where’s a Sony compact to rival the S95.