(DRAFT) VIDEO/TECHNOLOGY | Improving visibility, clarity of videos showing "talking demon heads"

I'm still writing; still putting together materials. Some videos I'm processing have over 10,000 still frames; and, some of those have at least three—if not more—demons in them...at all four angles. What's more is, in order to process them for improved visibility, you have to customize a procedure specifically adapted to the demon in the image. As already stated on this blog, most demons horde, and therefore, their features overlap.

One way demons distinguish themselves from each other visually is by shifting the color their reflect in order to differentiate themselves from their partner (what makes it hard to spot them is the way they alter individual qualities of color (hue, luma, chroma, saturation, lightness, value or what-have-you; while the hue of a given demon may differ from the demon next over, its lightness or saturation may match perfectly. At first glance, it's hard to tell them apart; and, even when it's easy, the differences between them completely alter the procedure for enhancements.

The other way they make a distinction in appearance is by differentiating their size. Sometimes, if you look closely at a giant demon head in an image, you will find it is made of three smaller ones, although to find them, you may have to rotate the image as many times.

That leads me to the third distinction: orientation. Each still frame must be rotated 360° in order to see all of the demons in an image. Given that fact, you have quadrupled the amount of work involved in skimming still frames for demons; at 10,000 with at least 3 demons, four ways (0°, 90°, etc.), equals 70,000 possibilities. It's hard to pick one, and it's hard to stop at even 70,000. They are as unique as snowflakes, and just as ornate in appearance.

I'll post images to demonstrate each method of distinction aforedescribed shortly; but, it's late, I'm injured, and would like to rest, if possible. Enjoy what I have so far, though...

The video this post will eventually talk about is still rendering, even as I sit here and write. Normally, I would wait until everything is ready to be published all at once before introducing a new post; but, occasionally, if the topic itself warrants immediate attention, as it does a lot with respect to demons, and if about half of what is ready to publish is sufficient to carry the topic, then I may decide even a half-post like this one will interest readers enough to visit the same post at least twice (once before it's ready, and again afterwards).

In this case, it was the still-frame rendering samples generated by the image-processing script I wrote to improve video showing the talking demon heads that pop-up during periods of high demonic activity on video. They are everywhere wherever I am, and I have tons of footage of these always stationary and mostly immobile talking demon heads, going back as far as three years ago.

Until now, have never found an occasion to post any of them for several reasons (one notable exception being my talking bedspread, as shown in VIDEO TIMELINE | Blanket Demon of My Youth, Face-to-Face).

For one, whenever they appear—which is always and only during a period of high demonic activity, they take a passive (inactive) role in the melee that ensues; they neither engage me or anyone (or anything) I talk to in conversation, nor seem take any interest in me or the going-ons. In fact, they seem quite oblivious to the world around them (although I think that's how demons avoid rudeness by the appearance of nosiness—no one will suspect or worry that you're listening in on their conversation or watching what they are doing if you act like you can't see or hear anything but yourself. Demons are very sensitive to social gaffes, of which there are a possible many, due to the fact that demons cluster [explain; link to post about shared consciousness in hordes], and privacy is at a premium as it is. I wasn't told that this is what it is, but it's the best educational guess one could make, and follows the same format for all the other rules demons have, but people don't.

Secondly, unless you're viewing a video showing one of them on the device that made it, you can't make them out very well; not only do they always appears in low-light settings, glowing dimly, but they are nearly always a recognizable pair of eyes, lips and a nose, while anything else goes.

Showing such videos to an audience with display devices with black points set anywhere but near yours, and with creatures in them they've never seen before, and could barely make out even if they have...

[mouth blurs on video due to EMF radiation emanating from demon; you can't hear them usually, unless you know how; the rate of speech is far, far faster than human normal, requiring a great deal of mental effort to hear and communicate (I usually fall asleep in about five minutes after talking to a demon that speaks like these talking heads do).

I accidentally ran the wrong script, and didn't notice until the first several still frames of the video were processed by it (it was a script used for another video, written about a week or so ago); however, surprisingly, I got better results with it than the script I wrote specifically for the video:

The first 8 still frames of the subject video, processed by a video filter intended for a completely different video, twice, looks far better than those processed by the video filter designed for the subject video

text.

Following are the original and the product of the script I wrote specifically for this video:

Original, unedited still frame from the subject video

Same still frame, processed by the video filter originally engineered for the subject video

The script originally written for the subject video:
#!/usr/bin/env python import gimpfu #access constants from gimpfu import * from os import listdir from os.path import isfile, join gettext.install("gimp20-python", gimp.locale_directory, unicode=True) def hilite(*args): mypaths = [args[0]] for mypath in mypaths: included_extensions = ['jpg','bmp','png','gif' ] ; images = [fn for fn in listdir(mypath) if any([fn.endswith(ext) for ext in included_extensions])]; for i in images: try: if i.lower().split(".")[-1] in ("png", "jpg"): image = pdb.file_png_load(mypath + i, mypath + i) # pdb.gimp_display_new(image) frame1 = pdb.gimp_image_get_active_layer(image) drawable = pdb.gimp_image_get_active_drawable(image) fill_layer = pdb.gimp_layer_new(image, pdb.gimp_image_width(image), pdb.gimp_image_height(image), 0, "fill", 100, 5) gimp.set_foreground((255, 255, 255)) pdb.gimp_drawable_fill(fill_layer, 0) value_layer = pdb.gimp_layer_new_from_visible(image, image, "value") pdb.gimp_image_add_layer(image, value_layer, 0) pdb.plug_in_normalize(image, value_layer) pdb.gimp_curves_spline(value_layer, 0, 6, [0,0,127,95,255,255]) pdb.gimp_desaturate_full(value_layer, 1) pdb.gimp_layer_set_mode(value_layer, VALUE_MODE) frame1 = pdb.gimp_image_get_active_layer(image) pdb.plug_in_wavelet_decompose(image, image.layers[0], 10, 3, 0, 0) pdb.gimp_layer_set_mode(image.layers[2], OVERLAY_MODE) pdb.gimp_layer_set_mode(image.layers[3], HARDLIGHT_MODE) pdb.gimp_item_set_visible(image.layers[1], FALSE) for x in range(4, 8): pdb.gimp_item_set_visible(image.layers[x], FALSE) pdb.gimp_layer_set_mode(image.layers[8], GRAIN_MERGE_MODE) pdb.gimp_layer_set_mode(image.layers[9], GRAIN_MERGE_MODE) pdb.gimp_item_set_visible(image.layers[10], FALSE) filename = pdb.gimp_image_get_filename(image) pdb.file_png_save_defaults(image, pdb.gimp_image_flatten(image), filename, filename) # pdb.gimp_displays_flush() pdb.gimp_image_delete(image) except Exception, error: print error break return

This is the script that I accidentally ran—twice—on the video, which produced (I think) better results:
#!/usr/bin/env python from gimpfu import * from os import listdir from os.path import isfile, join gettext.install("gimp20-python", gimp.locale_directory, unicode=True) def echo(*args): mypaths = [args[0]] for mypath in mypaths: included_extensions = ['jpg','bmp','png','gif' ] ; images = [fn for fn in listdir(mypath) if any([fn.endswith(ext) for ext in included_extensions])]; del layers[0] for i, j in zip(images, layers): try: if i.lower().split(".")[-1] in ("png", "jpg"): if j.lower().split(".")[-1] in ("png", "jpg"): image = pdb.file_png_load(mypath + i, mypath + i) # pdb.gimp_display_new(image) frame1 = pdb.gimp_image_get_active_layer(image) drawable = pdb.gimp_image_get_active_drawable(image) frame_overlay = pdb.gimp_layer_new_from_drawable(frame1, image) pdb.gimp_image_add_layer(image, frame_overlay, 0) pdb.gimp_layer_set_mode(frame_overlay, OVERLAY_MODE) pdb.gimp_invert(frame_overlay) pdb.gimp_desaturate_full(frame_overlay, 1) pdb.plug_in_gauss_iir2(image, frame_overlay, 15, 15) pdb.gimp_curves_spline(frame_overlay, 0, 6, [0,0,127,255,255,127]) pdb.plug_in_normalize(image, frame_overlay) pdb.plug_in_wavelet_decompose(image, image.layers[0], 10, 3, 0, 0) pdb.gimp_layer_set_mode(image.layers[2], OVERLAY_MODE) pdb.gimp_layer_set_mode(image.layers[3], HARDLIGHT_MODE) pdb.gimp_item_set_visible(image.layers[1], FALSE) for x in range(4, 8): pdb.gimp_item_set_visible(image.layers[x], FALSE) pdb.gimp_layer_set_mode(image.layers[8], GRAIN_MERGE_MODE) pdb.gimp_layer_set_mode(image.layers[9], GRAIN_MERGE_MODE) pdb.gimp_item_set_visible(image.layers[10], FALSE) filename = pdb.gimp_image_get_filename(image) pdb.file_png_save_defaults(image, pdb.gimp_image_flatten(image), filename, filename) # pdb.gimp_displays_flush() pdb.gimp_image_delete(image) except Exception, error: print error break return