Curiosity: Potential Anomalies (Update 01/2014)

reply to post by funbox


All.

But mostly the blue ones, as we can see the blue is applied to the whole block, even to the area in the background.

JPEG was made for compressing photos, but not to be zoomed-in.

ArMaP
That's an excellent example of how JPEG compression creates square blocks and applies the colours to the blocks to enhance the contrast.

That's why I try not to blow anything up. There is already going to be a lot of pixelation just from the very fact that we're looking at digital images though a digital screen. There's no way around that, but I like to try and minimize it, to the point where if I'm not sure about a particular shape I won't highlight it.

reply to post by ArMaP


that sharp v shape edge was not created by Jpg compression artefacts , their square , not triangular

eg



funBox

reply to post by Blue Shift


That's why your finds are usually very good, because of your non-zoom rule. The spokes you found, with no blow up, put them even further into their own category of possible marine/lake dwellers. Your barnacle is very good. Lots of other things you've found.

To funbox:

Unlike Blue Shift, I tend to look and zoom, as with the one I pointed out earlier which, on closed inspection, fell apart to individual rocks.



To everyone:

What could have made the jagged-edged thing Jeep pointed out? And about those lines in the sand which form squares and straight lined areas with almost no rocks in them, has anyone explained that as yet? thanks, and Char-Lee is missing, must be trampling around Mars and she'll be waving into the camera soon. She'll like the barnacle thing.

Aleister
reply to post by Blue Shift

 

To funbox:

Unlike Blue Shift, I tend to look and zoom, as with the one I pointed out earlier which, on closed inspection, fell apart to individual rocks.

plenty of valid information to be gained from zooming in , is it not so that nasa first download a shed load of thumbnails of varying quality in which some are picked to download a larger verion and some are not , these thumb nails show features but are major poor quality.. valid information in them ,, definatly.

funBox

funbox
that sharp v shape edge was not created by Jpg compression artefacts , their square , not triangular

I'm not talking about the triangular shape, obviously, I know the difference between a triangle and a square.

Don't you see all the squares in the brightened, zoomed-in image you posted, along with the blue squares on the areas close to the background, "bleeding" into the background? That's what I was talking about.

reply to post by ArMaP


indeed I was talking in geometry, you was busy looking at jpg artefacts, which was like you said further enhanced by the auto white balance I did on the blow up, although maybe a blue plaque build up afflicts marsharks teeth blue green algae mouth rinse maybe?



the images and the colours we are seeing from these raw (unprocessed) compressed jpg is just that unprocessed .
for true colour representation white balancing is needed, see pic for raw rgb histogram, they are far from synced/balanced



so , maybe the blue is there , and they are not artefacts of a compressive nature

funBox

reply to post by funbox


Could it be that some to-medium zoom works wonders (as the old saying goes), but "A Zoom Too Far" falls apart into pixelly chaos. Blue Shift's no-zoom policy finds give the best quality pics to point to when exobiologists show up on our doorstep, and other pics with some zoom don't fall far behind. Some others, with zooms of many hundreds of percent, seem harder to use as evidentially firm and focused logically nonsubliminal pieces of the puzzle when the exobio professionals and hired-hand exogeologists come calling and want to see the portfolio of planetary possibilities.

_____________________

And as funbox and ArMap discuss, over cognac and bluefish, the have or have nots of the little blue pyramid and the shiny blue flavorings, a stern Chief, solemn in continence and never blinking, as rocks tend to not do on occasion, stares out with a serious - verily, some would say with a worried - expression over the windblown unbarren landscape.

reply to post by Aleister


theres are a few factor to take into account , the landscape and how much of it is in frame would be a deciding factor , for instance
lets say the mastcam took a picture at the ground , then 100 % of the frame would be viable at a fixed level say 500% after that all areas would degrade roughly equally.

add distance and a horizon line.. usable zoom levels are going to drop off drastically the further away from the camera , and, in tandem the level of usable detail with it, add to this focus , lighting conditions , variable after variable ..sand storms for instance , they can crop up just on the horizon and obscure the mount almost completely ... so many variables

funBox

funbox
indeed I was talking in geometry, you was busy looking at jpg artefacts, which was like you said further enhanced by the auto white balance I did on the blow up, although maybe a blue plaque build up afflicts marsharks teeth

blue green algae mouth rinse maybe?

You were so busy thinking about geometry that you didn't understand that I was not talking about it.

for true colour representation white balancing is needed, see pic for raw rgb histogram, they are far from synced/balanced

I know, I have been doing this for years.

so , maybe the blue is there , and they are not artefacts of a compressive nature

Do you know how JPEG compression works? It really changes the colours to enhance contrast between colours, so we cannot really know how much blue is there on those rocks. From what I have seen in other photos, the blue is probably there, but I am sure is not in square patches that do not follow the shape of the objects, something that you didn't comment.

Aleister
Could it be that some to-medium zoom works wonders (as the old saying goes), but "A Zoom Too Far" falls apart into pixelly chaos.

The problem is resolution.
Regardless of being close to the camera or far away, if something (either an object or some detail of an object) appears in just two or three pixels then it's too small to get better with some zooming. That's why some of the photos that have been posted on this thread that show objects close to the camera, the "anomalies" are only on the smaller objects, those are the ones that have too little resolution to get a good idea of what they represent.

PS: on the HiRISE site, for example, when they say that the resolution of a photo is 0.25 metres per pixel, they also say something like "objects with more than 0.75 metres are defined", as they consider that we need at least 3 pixels to be sure that something is there.

reply to post by ArMaP


the color has to be there, for how would the pixel next door interpolate that information to the next pixel , or its guess work wouldn't be so messy in the compression process, when you take out the numbers and leave behind only a few to select from , what do you expect a pixel to do turn red, or a milder shade of blue ?

may I ask ArMaP , what is your main field of expertise? , Im figuring you have been educated to at least degree level

funBox

funbox
the color has to be there

No, and if you make some tests you will see that JPEG compression, sometimes, creates colour were there was none.

I saw once an image that had faint pink and green lines between dark and bright grey squares.

You can see it on these images.

First, a JPEG version


and the same image resized to 400% without resampling (and saved as a PNG to avoid more artefacts) to show the result of the JPEG compression.


The original image (also resized to 400% without resampling) was this one.


As you can see, the JPEG compression created colours that weren't there.

reply to post by ArMaP


now try that example with the actual image and say that the colors is not there never mind huge blocks of colour

hilarious

and as an afterthought , see the rock next door? did the color interpolate , at a rough guestimate 150 pixels ? I think not , the two rocks share the same colorations , and no amount of color bleeding off the edges and into the sand is going to change that

funBox

funbox
may I ask ArMaP , what is your main field of expertise? , Im figuring you have been educated to at least degree level

No special field of expertise or degree, but I have been working as a programmer for almost 20 years.

reply to post by funbox


If you don't see the cyan pixels near some of the corners of the red squares then I suppose it's no use talking to you about this subject.

funbox
reply to post by jeep3r

 

hey , where did the head one go ? that was a good one

, right next to it was an unusual patterning on the stone , see pic

You're right, I removed that 'funpic', but it's included in the above image and here's the source link again ...

ArMaP

funbox
the color has to be there

No, and if you make some tests you will see that JPEG compression, sometimes, creates colour were there was none.

I saw once an image that had faint pink and green lines between dark and bright grey squares.

You can see it on these images.

First, a JPEG version

and the same image resized to 400% without resampling (and saved as a PNG to avoid more artefacts) to show the result of the JPEG compression.

The original image (also resized to 400% without resampling) was this one.

As you can see, the JPEG compression created colours that weren't there.

Good examples, ArMaP ...

You mentioned the fact that resolution (and compression rates) are important. So taking into account the standard JPG compression in Curiosity's images, I could imagine that the color-bleeding effect mostly affects very small sections with only a few pixels (even more so when those are overly enlarged). The effect should IMO be nearly insignificant in most of the larger images used in this thread ... would you agree?

ArMaP

and the same image resized to 400% without resampling (and saved as a PNG to avoid more artefacts) to show the result of the JPEG compression.

As you can see, the JPEG compression created colours that weren't there.

More importantly it created objects in the picture, lines, blocks & squares TYPICAL of what we see in GROSSLY OVER ZOOMED images that many members post in these types of threads.

reply to post by wmd_2008


maybe you would like to example the original picture and show me what ArMaP failed to do on the original



white balanced



is the blue on this picture created by compression too ?



or this ?



how about the blue hues in the original blowup? , compression created?

whole picture white balanced



hmm seems like a lot of misconceptions are floating around about these colour photos

don't worry there's plenty of more blue stained rocks to example

and im not even going to use my pixel grid annoted original

funBox