The two weird and unexplained pictures show only one of the possible ways of making voltage and current sources - by using non-linear elements (without applying negative feedback). They are named "practical" in the sense "they are not ideal".
Voltage source
Static voltage divider. The simplest voltage source with desired constant voltage VREF can be made by a voltage divider supplied by a voltage source with constant voltage V > VREF. This humble electric circuit consists of two resistors (R1 and R2) in series. Both voltage drops VR1 and VR2 can be used as output voltage but, as a rule, the voltage drop VR2 across the grounded resistor (R2) is used. This voltage divider is "static" since its transfer ratio (gain) is constant - R2/(R1 + R2)... and this is the problem. If this was illustrated by another picture from this source, it would mystically say "the output voltage depends very much on the external parameters". For example, if V is not stable and varies, VREF will vary as well.
Dynamic voltage divider. The clever idea encrypted in the upper picture is to make the voltage divider dynamic. For this purpose, the "static" resistor R2 is replaced by a "dynamic" (self-variable) resistor R2dyn having a property to keep the voltage drop across itself constant. I am not sure what is its "official" name (if there is one at all) but it can be named with a descriptive name like "voltage-stabilizing dynamic resistor"... or "voltage-stable non-linear resistor"... or simply "voltage stabilizer"...
Operation. The dynamizing trick is simple and intuitive... and can be easily demonstrated by an ordinary variable resistor (rheostat). When, for some reason, the supply voltage V decreases, the current I decreases as well… and VREF (VOUT) should decrease. But the "clever" dynamic resistor R2 senses this change and increases its resistance to compensate the current decrease. As a result, VREF = Idec.R2inc = const. In other words, the voltage drop across R2dyn is a function of two oppositely varying input variables... and because of that it does not change.
Range. Dynamic resistors are passive devices (in the sense they do not produce energy). So their IV curve passes through the coordinate origin and can be roughly represented by two sections. The first is (almost) horizontal and not interesting for this application. The second becomes (almost) vertical "beyond Ith" (minimum current) and is useful for this application.
Implementation. Once revealed the concept, we can see many specific implementations of this idea by semiconductor devices having such a behavior - various diodes. So, the IV curve in the upper picture belongs to a diode.
Current source
Static current source. The simplest current source with desired constant current IREF can be made by a resistor R in series to a voltage source with constant voltage V. This current source is "static" since its internal resistance (R) is constant... and this is the problem. If this was illustrated by another picture from this source, it would mystically say "the output current depends very much on the external parameters". For example, if the supply voltage V or the load voltage VL… or both vary, IREF will vary as well.
Dynamic current source. The clever idea encrypted in the lower picture is to make the current source dynamic. For this purpose, the "static" resistor R is replaced by a "dynamic" (self-variable) resistor Rdyn having a property to keep the current through itself constant. Again, I am not sure what is its "official" name (if there is one at all) but it can be named with a descriptive name like "current-stabilizing dynamic resistor"... or "current-stable non-linear resistor"... or simply "current stabilizer"...
Operation. The dynamizing trick is also simple and intuitive... and can be easily demonstrated by a variable resistor (rheostat). When, for some reason, the supply voltage V decreases or the load voltage VL increases, the effective "current-creating voltage" across Rdyn decreases… and IREF (IOUT) should decrease. But the "clever" dynamic resistor Rdyn senses this change and decreases its resistance to compensate the current decrease. As a result, IOUT = Veffdec/Rdyndec = const. In other words, the output current IREF is a function of both input variables (voltage and resistance) that changes in the same direction... and because of that their ratio - the current, does not change.
Range. These dynamic resistors are also passive devices (they do not produce energy). So their IV curve passes through the coordinate origin and can be roughly represented by two sections. Now the first is (almost) vertical and not interesting for this application. The second becomes (almost) horizontal "beyond Vth" (minimum voltage) and is useful for this application.
Implementation. Once revealed the second concept, we can see many specific implementations of this idea by semiconductor devices having such a behavior - now various transistors. So, the IV curve in the lower picture belongs to a transistor (it is the so-called "output characteristic").
Finally, let me just add that when I come across such "explanations", I feel happy that I was not forced to take an exam in such lectures...