Home   All Snags in the Cosmos   Previous   Arch.aeology and Photographic Dinosaurs

Artefacts & Crackpot Bones Cocorico !

 
PDF File
 

  1. Preliminary note:

There are "artefacts" and "artefacts":

  1. First those which are generated by the data processing itself.

  2. Then those that are already included in the image before this processing, and that the treatment reveals.

  3. And those that are generated by the files compression realized by observatories before their publications.

Let us examine some examples of these three cases, to obtain some useful conclusions.

  1. Artefacts due to the data processing:

No real available example to this day. (09/15/2000).

 



 

Indeed, every time seemingly strange things are revealed, it appears, in every case, that it is about objects or about very real phenomena, connected with the photographed objects. We always find a link with the original photo and often a rational explanation.
But sometimes this explanation is "unconventional". It is a disturbing fact. But the facts are there. And as we do not work to please or to displease anyone, we admit the facts such as they are.
The example of artefact which we present below, was obtained by the "Curves" method with Adobe Photoshop. It reveals a very real physical phenomenon in the galaxy M87.
In spite of appearances, It is not an artefact. It shows that the method can, in certain cases, be used to estimate the gradient of luminosity of the gas, and thus the gradient of density of the gas in a jet of plasma, in a nebula or in a galaxy...


 
 
  1. Artefacts contained in the image:

There we have numerous examples.

It is very surprising to see that certain images retain "instrumental marks" which allow identification of their origin! And especially the methods which presided over their making.
The examples, shown below, are almost exaggerated. But they validate, by repercussion of their existence, the used method. They are deliberately chosen of different origin, to show that these artefacts are not the privilege of any particular source.



 

1st example: NGC 253 (VLT)

The processing of the image of this galaxy shows, not only the real dimensions, but also quite a lot of incongruous things. First two diagonal lines which are, after information was requested from an official of ESO, traces left by the passage of one or two artificial satellite(s) during the exposure. But what is even more extraordinary, it is the grid pattern that appears and which is the archaeological track of the making of this photography.


 



 

2nd example: Virgo Cluster (NOAO)

This image presents, after processing, numerous abnormalities. The most striking appear in the zone located by the rectangle. It is obviously a transparent label, with a text, stuck on the back of the photographic plate. The digitalization was made without too many precautions, without recentering correctly. To decipher this text it is necessary to look at it by using a mirror. The word which is manifestly registered is "NORTH". We understand easily the reason of its presence!


 

  1. Artefacts brought by the computer compression of the files of images:

These artefacts are also included in the images. They are brought by intermediate processings between the recording and the creation of the final numeric image. In particular by the techniques of compression.
The most common case is the compression to the format JPG. Greater the compression ratio is, then more important these artefacts are.
The attempts which we realized on a TIFF - RVB reference image of JPG format, created for this purpose, and having the following characteristics: dimensions 898 x 585 pixels, 41330 bytes, are very revealing. We converted this image to the JPG format with compression ratios ranging from 0 to 75 %. On every obtained image, we did the previously described processing. Which accentuated the artefacts, without degrading them.
The table below gives the obtained results:

 

Before Processing

After Processing

Comments

TIFF

41330 octets

56668 octets

There is no perceptible modification of the quality of the image, even with a strong magnification.

       

JPG

0 %

28488 octets

47406 octets

The quality of the JPG image  is not degraded in a significant way, The processed image does not show perceptible degradation.

       

JPG

25 %

15061 octets

59149 octets

With an trained eye, the JPG image shows very light degradations. But not to the point of generating artefacts. The processed image shows these same degradations but with added contrast.

       

JPG

50 %

13289 octets

56491 octets

With an trained eye, the degradations of the image are a little clearer. But afterprocessing they are indeed revealed.

       

JPG

75 %

11922 octets

52801 octets

The degradationss are unacceptable, and the processed image makes them more perceptible.



 

Here is an example of the degradations brought by the compression JPG to 50 %. in the central region of both images. To the left the image JPG and below the corresponding processed image JPG.

The structure of the degradations is the same in both cases. But it is revealed very well in the processed image.

This structure is very characteristic of the JPG compression. We are in the presence of a real signature which cannot deceive. (Appearance of oblong zones stained with pixels of densities and with variable colours).

These artefacts are present only in the zones of transitions of colours or luminosity. Their sizes are limited, according to the compression ratio.

In other words, in a large image, and if there are few zones of transitions, these artefacts are not necessarily annoying. Thus, with the same compression ratio, an image of 1024 x 768 pixels can be completely acceptable, while an image of 100 x 100 pixels can be useless. So a preliminary exam of the image to be processed is indispensable to reveal this kind of signature.


 

Source NOAO (a102.jpg)

 

Here is a real example of compressed images made from a NOAO picture (galaxies Cluster Arp 102).

Extrait de a102.jpg

 

Both images opposite are strongly enlarged extracts of the zone at the end of the galactic right arm in the NOAO image above.

The opposite image at the top is obviously perfectly useful for making a computer processing, but that of the bottom one is absolutely not. Let us note here that these artefacts have nothing in common with what was discovered in the core of the galaxy M 87. In particular we do not see any tore or jet.

Extrait de a102b.jpg

 

CONCLUSIONS

  1. This method of extraction has its limits. We cannot use a too strongly compressed JPG image. 50 % is really a maximum. The simple conversion from TIFF to JPG without compression allows to obtaining images of reduced sizes, without altering the image.

  2. It is imperative to examine attentively the image we want to process to make sure that artefacts of compression are not present.

  3. This method does not generate, by itself, real artefacts. It sometimes reveals some that already are included in the original photos.

  4. The obtained images can present problems of interpretation. The example of M 87 with the curves of equidensity is a characteristic example. All the processed images can present this phenomenon to a certain extent.

  5. It is nevertheless a powerful tool for revealing unexpected phenomena, but with the prerequisite of knowing how to interpret the results.

Documents

  1. What's in a picture? The temptation of image manipulation. This excellent document shows how fraudulent manipulations can be revealed, simply by generating artefacts. However, the reader will notice that a digital processing realized cautiously is perfectly justifiable, especially if it aims to emphasize some phenomenons which otherwise would remain invisible.

  2. Images et graphiques pour le web - Jean-Claude SOHM - CERIG / EFPG (11/28/2003).

  3. Forum des Sceptiques du Québec - la cosmologie révisée par Photoshop. We never intervened directly in this forum. Opinions often very critical on the method described here, were expressed. We took into account it to improve this method and to avoid certain traps. We thank all the contributors and more particularly one of them who keeps us informed.

 

Last Update: 09.14.2014

Return:

Home