Interesting Phoenix image

[RAF]

Any ideas as to what the white "object" in the background of this image might be??

Part of the parachute assembly/aeroshell perhaps??

[Data Cable]

<rimmer>Aliens!</rimmer>

[RAF]

May 25, 2008 23:40:00 GMT -4 Data Cable said:

<rimmer>Aliens!</rimmer>

Yeah...I'm sure hoagwash will momentarily post his usual junk on his site.

Opinions on the web are favoring the heat shield as the most likely explanation.

Can't wait to see some HIRES images of the whatever it is.

[PhantomWolf]

It'd be interesting to see if it is actually in other images. Part of it looks like a processing or transmission error to me.

[Ginnie]

You guys should be able to figure it out by now.

It's a water cooling tower

[Czero 101]

There's also this one...



which I think shows the aeroshell. Just wish there was a larger, more detailed pic available.

ETA: I think the same object appears in the top-left corner of the image area of this pic:


Click image above for high res image (1733x1192, 140kb)

Cz

[JayUtah]

Single-bit transmission errors. Single bit errors in the wrong place can really mess up one or more DCT fields. The DCT fields in question look like they were encoded with only one or two elements.

[JayUtah]

Well, the second photo can't be attributed to a transmission error. But I can't verify same position. Did the rover move between photos?

[Czero 101]

Its not a rover, Jay... just a lander, along the lines of the Viking landers of the 70's:


Click to link to higher res image 2395x1499, 640kb


Cz

[JayUtah]

So I guess that would be "no."

[Czero 101]

Yes.... no...


;D


Cz

[raven]

Hey lets all play 'Spot the Gestalt Theory Example Weird Rock"
Just above -20 on the left edge of the image, I spy something that looks a LOT like a head. Anyone else spot any "anomolies"?

[RAF]

I agree that the image I posted in the OP could be some form of transmission error, and I also agree that the image posted by czero could very well be the aeroshell...

Boy, am I an "agreeable gus" or what?

[Jason]

An elongated crystaline skull. Probably also magnetic.

[JayUtah]

Notice how the believers in magnetic crystalline skulls rather quickly acquire the habit of babbling incoherently?

With only one photograph in hand -- the first one posted -- the nature of the anomaly alone suggests single-bit error. In JFIF-encoded DCT transformations, some bits are more important than others, which is to say that some bits have more far-reaching consequences than others and thus have higher criticality. The bits, for example, that set the baseline intensity for a field are important because the relative departures therefrom, as given in the frequency-domain components, will fail to correct it back to the expected intensity. In more understandable terms, it's like surveying: if your starting point was wrong then it doesn't matter how precisely you follow the relative distances and directions -- they will all subsequently identify the wrong points.

One DCT field in the anomaly has an entire column of pure white pixels. The remainder of the field exhibits pixels consistent with low-order DCT approximations. The low-order behavior is appropriate: several other non-anomalous fields also exhibit low-orider approximation, but their intensities are generally consistent with the surrounding fields and are therefore not immediately interpreted as having a high likely degree of loss. This is why the JPEG group's engineering is well founded and has proven so successful.

However, a single-bit error in the baseline intensity for a field will cause a low-order approximation in that field to announce itself by field-wide variation. The low-order approximation produces an entire field of seemingly anomalous pixels.

But with more data in hand, the transmission error hypothesis becomes more credible if the feature correlates to objects in the landscape between photographs. That is, multiple photographs showing a salient feature are less likely to be encoding, compression, and transmission artifacts.

We further note that the first data set contains two DCT fields with exceptionally bright pixels. Originally I believed this was consistent with the way baseline field intensity values are themselves DCT-encoded across the entire image, but on consulting the nuts and bolts of the baseline encoding algorithm I find it to be fairly improbable that two fields diagonally related in upper-left, lower-right fashion will be affected by a common single-bit error; they are actually separated in the image-wise signal by a significant distance.

This is not a smoking-gun rule-out, however, because in practice telemetry streams reorder the data prior to modulation such that a multiple-bit data drop-out will not desynchronize the framing and blocking algorithms. That is, bits that start out adjacent in the pre-modulated data stream will not be adjacent in the time domain of the modulated signal. This allows the possibility that a two- or three-bit error in the modulated signal will produce two or more widely spaced errors in the demodulated data stream. However, I don't have enough data to compute the probability of a tw-bit error (pun intended) producing a widely-spaced error in the DCT baseline data set. That is, it's easy to believe that a transmission glitch can produce two separated data errors, but the chances of them falling simultaneously on baseline DCT data seems remote.

It is also possible that low-order DCT compression can produce field-wide salient results if the field contains high-frequency, high-amplitude variation that cannot be accommodated in a low-order waveform. This will require the relatively invariant portion of the field to accept high-amplitude variation in order to provide mathematical leeway to best-fit the bright feature in a low-order wave.

If the location of the bright feature in the first data set can be correlated in scene space with the feature tentatively identified as the aeroshell in other photographs, then the parsimonious explanation is that the bright feature at top is an actual feature that has affected two DCT fields.