< Mirosław Kordos: The best artificial intelligence solutions.


Introduction

As you have probably noticed: most musical recordings do not sound great, and most photos do not look amazing? Not the artists are to blame for this, but the ancient technology that we still use in 2023, although it currently does not make any sense.



How to prepare photos

Static file formats, as: jpeg, webp, heif or mp4 belong to the past and should no longer be used nowadays. It's time for dynamic (inactive) media formats. Let me explain this. 

When you prepare a photo for publication, several decisions have to be made, of which the most problematic ones are the amount of saturation, the color temperature, and extension or contraction of dynamic range. But how can you do it, if you don't know on which kind of display this photo will be looked at? Should you prepare it for a good monitor like Asus Proart PA32UCR or PA32DC? And if so in which display mode? Adobe RBG, DEC 2000, DPI-P3, or another mode? Well, most people do not have good monitors, so you can prepare the photo for a popular cheap displays. But is it a good solution? Someone who has a good monitor can appreciate your photo, and someone who has not, frequently does not pay any attention to the photo quality. 

The reasonable solution may seem to prepare two versions. One for good monitors for one of the wider gammut modes, one for popular cheap displays, and of course you must write what a given version of the photo is optimized for.

But what if someone has a monitor that cannot be classified into either of the two types? For example, the Gigabyte M32U. It is a monitor with a quantum dot technology, so the colors it displays are more vivid and saturated than those on standard monitors, and also the gammut is wider. However, it is by no means as good as the monitors designed for graphics, as Asus Proart PA32UCR or, even better Asus Proart PA32DC. So what now? The third version of the photo? And what about PVA displays? The fourth version? By the way, the two mentioned Asus models are also different - PA32DC is an OLED display with better blacks. OK - that gives five versions of the photo. Surely, this is not the way to go.

The way to go is a dynamic format, which has a metadata section, which contains parametrized information about the saturation, dynamic range, sharpness, temperature, etc. in each area of the photo and about the gradients. Then the proper system driver provides the information of the display to all the programs in which you can open the photo (or the display firmware does it), and the optimal parameters are automatically selected. But of course, the user must have the possiblity to override them. It is so simple! We have the 2023 year! I really do not understand how it happened that this is not in common usage!?!



How to prepare music

Static file formats, as: wave, flac, mp3 or mp4 belong to the past and should no longer be used nowadays. It's time for dynamic (inactive) media formats. Let me explain this. 

When you prepare a piece of music for publishing, several decisions have to be made, of which the most problematic ones are the amount of equalization (proportions of bass and trebles in different bands), the dynamic range (both temporal and global), the amount and loudness of details, the amount of reverb, and channel separation. But how can you do that, if you don't know on what kind of equipment and in what environment will this music be listened to? Should you prepare it for a good equipment like Proac Response loudspeakers or Martin Logan electostats? Or maybe for good headphones like Sennheiser HD600 with good headphone amplifier? (You may argue if HD600s are the best, but surely are rather good than bed haedphones, and are probably the most popular ones in the world, so it makes sense optimize rather for them, than for a model you may find superior, but rarely used.) Or maybe for someone who listens to it on a smartphone with cheap headphones in a noisy environment? Or maybe for some open-air festival or for a party in a big ballroom?    

Is the reasonable solution to prepare a dozen versions? One for Proac Response, one for Martin Logan Electromotion, one for a smartphone, ....., etc. Of course, you must write what a given version of the song is optimized for. 

The way to go is a dynamic format, which has a metadata section, which contains parametrized information about the channel separation (smaller for headphones, bigger for loudspeakers), reverb (this time bigger for headphones, smaller for loudspeakers, and frequently no reverb at all, as the natural reverb of the listening room does the job), dynamic range (temporal must always be big, but global dynamic range is another story), and equalization, which of course depends on the listening condition. To put it simply, the proper system driver or firmware of the DAC should provide that information of the connected equipment and listening conditions and modify the metaparameters accordingly to provide the best listening experience (but the user needs to be given the possibility to override the parameters). It is so simple! We have the 2023 year! I really do not understand how it happened that this is not in common usage!?!



And what about movies and video production?

The way to go is a also a dynamic format, where the matadata contains both the audio and video parameters.






Creative Commons License. You are free to copy, share and adapt all articles and software from my web page, provided that you attribute my works and place a link to my home page. What you build upon my works may be distributed only under the same or similar license, and you may not distort the meaning of my original texts.